You are on page 1of 85

ASPECTS OF PRECISION

AGRICULTURE
Francis J. Pierce' and Peter Nowak*
'Department of Crop and Soil Sciences
Michigan State University
East Lansing, Michigan 48824

2Department of Rural Sociology


University of Wisconsin
Madison, Wisconsin 53706

I. Introduction
A. Definition of Precision Agriculture
B. Intuitive Appeal
11. Overview of the Basic Components of Precision Fanning
A. The Enabling Technologies
B. Steps in Precision Agriculture
HI. Conclusions
References

Precision agriculture is the application of technologies and principles to manage


spatial and temporal variabilityassociated with all aspects of agricultural production for
the purpose of improvingcrop performance and environmental quality. Success in pre-
cision agriculture is related to how well it can be applied to assess, manage, and evalu-
ate the space-time continuum in crop production. This theme is used here to assess the
current and potential capabilities of precision agriculture. Precision agriculture is tech-
nology enabled. It is through the integration of specific technologies that the potential
is created to assess and manage variability at levels of detail never before obtainable
and, when done correctly,at levels of quality never before achieved.The agronomic fea-
sibility of precision agriculture has been intuitive, depending largely on the application
of traditional management recommendations at finer scales, although new approaches
are appearing. The agronomic success of precision agriculture has been limited and in-
consistent although quite convincing in some cases, such as N management in sugar
beet (Bera vulgaris L.).Our analysis suggests prospects for current precision manage-
ment increase as the degree of spatial dependence increases, but the degree of difficul-
ty in achieving precision management increases with temporal variance. Thus, man-
agement parameters with high spatial dependence and low temporal variance (e.g.,

I
A&mctx in Agronomy, Volume 67
Copyright8 1999 by Academic Prea. All rights of reproduction in m y farm reserved.
0065-21 13/99 $30.00
2 FRANCIS J. PIERCE AND PETER NOWAK

liming, P, and K)will be more easily managed precisely than those with large temporal
variance (e.g.. mobile insects). The potential for economic, environmental, and social
benefits of precision agriculture is complex and largely unrealized because the space-
time continuum of crop production has not been adequately addressed.
0 1999 Academic Press

I. INTRODUCTION

It would be a simple matter to describe the earth’s surface if it were the same
everywhere. The environment,however, is not like that: there is almost endless
variety.
-Webster and Oliver ( I 990)
The quote by Webster and Oliver (1990) is particularly applicable because pre-
cision agriculture is concerned with the management of variability in the dimen-
sions of both space and time. Without variability, the concept of precision agri-
culture has little meaning (Mulla and Schepers, 1997) and would never have
evolved. It appears that any component of production agriculture-from natural
resources to plants, production inputs, farm machinery, and farm operators-that
is variable in some way is included in the realm of precision agriculture. Aspects
of precision agriculture, therefore, encompass a broad array of topics, including
variability of the soil resource base, weather, plant genetics, crop diversity, ma-
chinery performance, and most physical, chemical, and biological inputs used in
the production of a crop, whether natural or synthetic. By necessity, these aspects
are all framed within the context of the socioeconomicaspects of production agri-
culture because to be successful on the farm, precision agriculture must fit the
needs and capabilities of the farmer (Nowak, 1997) and must be profitable (Low-
enberg-DeBoer and Swinton, 1997).
Bell et al. (1995) state correctly that efforts toward precision agricultural man-
agement should recognize that the factors affecting crop yields and environmen-
tal sensitivity vary in both space and time. Managing soils and crops in space and
time is the sustainable management principle for the twenty-first century, a prin-
ciple exemplified by farming by soilscapes, managing zones within the field, and
managing the noncrop period (Pierce and Lal, 1991). The unifying theme of this
chapter is that success in precision agriculture is directly related to how well it can
be applied to manage the space-time continuum in crop production. We postulate
that prospects for precision management increase as the degree of spatial depen-
dence increases, but the degree of difficulty in achieving precision management
increases with temporal variance. Thus, for management parameters that vary spa-
tially, those with high temporal correlations (e.g., liming) will be more easily man-
ASPECTS OF PRECISION AGRICULTURE 3

aged with precision agriculture than those with large temporal variance (e.g., mo-
bile insects). Within a given management parameter, the success to date of preci-
sion management is to a large extent determined by the degree to which the spa-
tial variability is temporally stable.
This chapter provides an overview of precision agriculture and an assessment
of its current state and its potential to improve crop performance and environ-
mental quality in production agriculture. In this chapter, we define precision agri-
culture, explore the technological capabilities that enable it, assess its agronomic
feasibility and environmental efficacy, and evaluate its performance to date rela-
tive to economic and social impacts. The chapter concludes with an analysis to
identify needed developments in precision agriculture and we provide some
thoughts for a future research agenda. Given the expansive nature of precision
agriculture coupled with space constraints, we attempt to synthesize the important
aspects of precision agriculture while guiding the reader to the growing volume of
literature on the subject. Readers seeking more detail are referred to the following
major publications related to precision agriculture: Auernhammer ( 1994),Ameri-
can Society of Agricultural Engineers (ASAE) (1991), BIOS (1997), Lake et al.
(1997), National Research Council (NRC) (1997), Pierce and Sadler (1997),
Robert et al. (1993, 1995, 1996), Sawyer (1994, Schueller (1992), Stafford
(1996b), and Sudduth (1998). We are aware of the rapid rate of change in preci-
sion agriculture and the inadequacies this causes in an overview of this nature.

OF PRECISION
A. DEFINITION AGRICULTURE
Currently, no precision agricultural systems exist; rather, various components
of traditional crop management systems have been addressed separately regarding
their potential for site-specific management, perhaps most notably soil fertility
(Lowenberg-DeBoer and Swinton, 1997).The state of precision agriculture from
a systems perspective is analogous to the early days of no-tillage crop production.
Technology became available in the 1960sto plant seeds in untilled soil, but it was
not until the many aspects of crop production were adequately addressed under
lack of tillage and crop residue management, including the management of fertil-
ity and pests, that successful no-tillage systems were developed and implemented
(Blevins et al., 1998). The adoption of no-tillage did not proceed at a significant
rate until the 1980s when the integration of appropriate technologies and public
policies supported its dissemination to farmers (Allmaras et al., 1998; Larson et
al., 1998;Now& and Korsching, 1998).In a similar fashion, although certain tech-
nologies in the early days of precision agriculture allowed for the variable appli-
cation of nutrients and pesticides, there did not exist a thorough understanding of
how soil fertility and pests varied in space and time. Most important, explanations
were lacking on what specifically caused the variability so that appropriate inputs
4 FRANCIS J. PIERCE AND PETER NOWAK

could be matched to site-specific conditions. Today, farmers are adopting individ-


ual components of precision agriculture on the farm but a distinctive precision
farming system has not yet evolved. Technological developments continue to oc-
cur and as a result of ongoing research a better understanding of underlying
processes is being developed but a true system has not emerged. Therefore, any
definition of precision agriculture can at best be considered only operational.
Since the mid-l980s, a host of terms have been used to describe the concept of
precision agriculture, including farming by the foot (Reichenberger and Russ-
nogle, 1989), farming by soil (Carr et al., 1991; Larson and Robert, 1991). vari-
able rate technology (VRT) (Sawyer, 1994), spatially variable, precision, pre-
scription, or site-specific crop production (Schueller, 1991), and site-specific
management (Pierce and Sadler, 1997). All these terms, however, have in com-
mon the concept of managing variability at scales that are within fields. As
Stafford (1996b; p. 595) states, precision agriculture involves “the targeting of in-
puts to arable crop production according to crop requirements on a localized ba-
sis.” Thus, the intent of precision agriculture is to match agricultural inputs and
practices to localized conditions within a field to do the right thing, in the right
place, at the right time, and in the right way (Pierce ef al., 1994). A recent report
of a National Research Council, Board on Agriculture Committee defined preci-
sion agriculture as “a management strategy that uses information technologies to
bring data from multiple sources to bear on decisions associated with crop pro-
duction” (NRC, 1997; p. 17). While the NRC definition raises important infor-
mational dimensions of precision agriculture, it fails to emphasize the basic
premise of precision agriculture-the management of spatial and temporal vari-
ability. In this chapter, we use the following definition of precision agriculture as
the basis of our discussions: Precision agriculture is the application of technolo-
gies and principles to manage spatial and temporal variability associated with all
aspects of agricultural production for the purpose of improving crop performance
and environmental quality.
We provide a final note on the word precision because there is sure to be con-
fusion regarding its meaning in precision agriculture versus its use in statistics. The
term precision refers to the quality or state of being precise, where precise means
minutely exact, a term synonymous with correct. Precision agriculturerefers to ex-
actness and implies correctness or accuracy in any aspect of production. In statis-
tics, however, precision is the closeness of repeated measurements of the same
quantity to each other, whereas accuracy is the closeness of a measured or com-
puted value to its true value (Sokal and Rohlf, 1995). In measurements, accuracy
is synonymous with correctness (i.e., validity), whereas precision refers to repro-
ducibility (i.e., reliability). Thus, something can be precise but not accurate. An-
other matter is measurement precision implied by number of digits reported for a
given measurement. The nature of computers makes it easy to imply more preci-
sion than was possible in various aspects of data collection, analysis, and compu-
ASPECTS OF PRECISION AGRICULTURE 5

tation in precision agriculture. Precision here refers to the limits on the measure-
ment scale between which the true measurement is believed to lie, implied by the
number of digits reported for a measurement (Sokal and Rohlf, 1995). The more
digits reported for a measurement, the higher the precision implied. A pH of 5.44
implies more precision than a pH of 5.4. The appropriate precision with which to
report a number is to include one additional digit beyond the last significant one
measured by the observer. Statistics plays an important role in the application of
precision agriculture and care should be taken in dealing with accuracy, precision,
and implied precision in the reporting data.

B. INTUITIVEAPPEAL
Precision agriculture is intuitively appealing because it is closely aligned with
the scientific principles of management of soils, crops, and pests. Few would ar-
gue against a management philosophy that espouses matching inputs to the exact
needs everywhere. Precision agriculture is intuitively appealing because it offers
a means to improve crop performance and environmental quality in production
agriculture (Wolf and Nowak, 1995). While the intuitive appeal creates high ex-
pectations for precision agriculture, the physical evidence supporting the agro-
nomic (Lowenberg-DeBoer and Swinton, 1997; Sawyer, 1994) and environmen-
tal (Larson et al., 1997) benefits of precision agriculture is limited in part because
it is still in its infancy.
As we will demonstrate in our discussion, successful implementation of preci-
sion agriculture depends on numerous factors, including (i) the extent to which
conditions within a field are known and manageable, (ii) the adequacy of input rec-
ommendations, (iii) the degree of application control, and (iv) the degree of sup-
port through private and public infrastructures. Individual success also depends on
the expectationsplaced on precision agriculture which represent the difference be-
tween promotional and educational efforts versus the actual experience of farmers.

II. OVERVIEW OF THE BASIC COMPONENTS


OF PRECISION FARMING

The main componentsof any precision agriculture system that may emerge must
first address the measurement and understanding of variability. Next, this system
must use information to manage this variability by matching inputs to conditions
within fields using site-specific management recommendations and mechanisms
to control the accuracy of site-specific inputs. Finally, and most important, this sys-
tem must provide for the measurement and recording of the efficiency and effica-
6 FRANCIS J. PIERCE AND PETER NOWAK

cy of these site-specific practices in order to assess value on and off the farm. Thus,
precision agriculture is technology enabled, information based, and decision fo-
cused (Pierce, 1997a).

A. THEENABLING
TECHNOLOGIES
While the concept of matching inputs to site-specific conditions is not new, as
just discussed, there is little doubt that important advances in technology contin-
ue to enable precision agriculture. The enabling technologies of precision agricul-
ture can be grouped into five major categories: computers, global position system
(GPS), geographic information systems (GIS), sensors, and application control.
Few of the enabling technologies were developed specifically for agriculture and
their origins date back more than 20 years, as illustrated in the time chart in Fig-
ure 1. It is the integration of these technologies that has enabled farmers and their
service providers to do things not previously possible, at levels of detail never
before obtainable, and, when done correctly, at levels of quality never before
achieved (Fortin and Pierce, 1998).

1. Computers

Many technologies support precision agriculture, but none is more important


than computers in making precision agriculture possible. Also, it is not computers
alone that are important but their ability to communicate that is so powerful for
agriculture.As Taylor and Wacker (1997) suggest, it is the fusion of computers and
communication that gave birth to connectivity, and it is connectivity that is driving
the access of everyone to everyone, everything to everything, and everything to
everyone. This electronic linkage and communication define the age of access
(Taylor and Wacker, 1997). It is this notion that may have prompted the NRC
(1997) to define precision agriculture in terms of a management strategy that uses
information technologies for decision making.
Precision agriculture requires the acquisition, management, analysis, and out-
put of large amounts of spatial and temporal data. Mobile computing systems were
needed to function on the go in farming operations because desktop systems in the
farm office were not sufficient. These mobile systems needed microprocessorsthat
could operate at speeds of millions of instructions per second (MIPS), had expan-
sive memory, and could store massive amounts of data. The first microchip creat-
ed by Intel in 1971 (Intel 4044 processor) contained a mere 2300 transistors and
performed about 60,000instructions per second. Since 1971, the number of tran-
sistors per chip has doubled every 18 months (Fig. 2) affirming Gordon Moore’s
observation in 1965 that a doubling of transistor density on a manufactured die
was occurring every year, a concept referred to as “Moore’s law” (Moore, 1997).
Date Event

1840s Aerial photography emerges; pictures taken from balloons


1960s First image sensors incorporated in satellites; low-resolution black and white TV
First commercial CIS
1961 First chlorophyll sensor (Benedict and Swidler, 1961)
1968 First multispectral photography done from space Apollo 9 manned mission
1970 Baumgardener et al. (1970) related soil organic matter to multispectral data
1971 Intel 4040 processor
1972 Launch of Earth Resources Technology Satellite-1 later renamed Landsat; permitted
continuous coverage of most of the earth's surface
1974 Soil organic matter sensor (Page, 1974)
1977 Apple computer commercialized (http://www.apple.com)

1978 Launch of first NAVSTAR GPS satellite


1980 First IBM PC
1981
1982 Intel 80286 processor
Launch of Landsat "Thematic Mapper ("h4) added
The Jet Propulsion Lab produces hyperspectral sensors for use from a high-altitude
aircraft platforms known as AIS (Airborne Imaging Spectrometer)
1983 GPS available for civilian use
1984 Ortlip patent issued to SoilTEQ
Launch of Landsat 5
286 Intel processor
1985 Grain flow monitoring on combines @e Baerdwmeker et al., 1985)
1986 French launch an operational series of earth-observingsatellites called SPOT (SysPme
Probatoire d'observation de la Terre); first offering of multispectral data to world
users on a commercial basis
1987 The Jet Propulsion Lab produces a second hyperspectral sensor known as AVIRIS
(Airborne VisibldnfraRed Imaging Spectrometer)
Yield mapping in Texas (Bae et al.. 1987)
1988 India launches earth resources satellite (IRS-IA) that gathers data in the visible and
near IR with the Linear Imaging Self-scanning sensor (LISS)
Intel 40486 processor
Canadian Radarsat, ERS-I, and ERS-2 managed by the European Space Agency A
class of satellite remote sensors using radar systems
I990 Japan launches JERS- 1 and JERS-2 that include both optical and radar sensors
Selective availability (SA) imposed on GPS signal
1991 First symposium on site-specific crop production (ASAE, 1991)
1992 Commercial yield monitors appear in the United States
First international conference on soil specific crop management (Robert et al., 1993)
1993 Pentium processor
I994 Full constellation of 24 GPS satellites in NAVSTAR system complete
1996 Earth System Science Pathfinder launched by NASA
1997 Pentium I1 processor
India launches the latest in the series. IRS-ID, on September 29, 1997
First European conference on precision agriculture (BIOS, 1997)
Board of Agriculture, National Research Council report on precision agriculture
(NRC,1997)

Figure 1 Historical developments in the technologies that enabled precision agriculture.


8 FRANCIS J. PIERCE AND PETER NOWAK

Yorr
Figure 2 Illustration of Moore’s law showing the doubling of computer speed and capacity every
year [Source:Intel Corporation (www.intel.com)].

As Fig. 2 indicates, Moore’s law is expected to hold until 2017 (according to


Moore) and appears to hold for memory and storage capacity. Data storage ca-
pacity will need to increase rapidly as sensor technology and digital geospatial data
become increasingly available to agriculture. Moore notes,
By the Year 2012, Intel should have the ability to integrate 1 billion transistors
onto a production die that will be operating at 10 GHz. This could result in a
performance of 100,OOOMIPS, the same increase over the currently cutting edge
Pentium I1 processor as the Pentium 11processor was to the 386! We see no fun-
damental barriers in our path to Micro 2012, and it’s not until the Year 2017 that
we see the physical limitations of wafer fabrication technology being reached.
We can expect, therefore, that computers will drive significant technological de-
velopment to enable precision agriculture for the foreseeable future. The extent to
which agriculture can utilize computer technology is important to the success of
agriculture in general (Holt, 1985; Ortmann et al., 1994). However, the agricul-
tural sector is lagging in the adoption of computer technologies on the farm rela-
tive to other business sectors. According to the 1997 annual survey of the U.S. De-
partment of Agriculture (USDA) National Agricultural Statistics Service (NASS,
1997), of 2,053,800 farms in the United States, only 38% had computer access,
31% owned or leased a computer, and 13% had Internet access. Part of this com-
puter lag in agriculture is due to the lack of access or connectivity in rural areas,
lack of training, and little perceived utility in available software (Nowak, 1997;
ASPECTS OF PRECISION AGRICULTURE 9

Peterson and Beck, 1997).In any event, farmers will have to become as comfort-
able working with computers and their data as they are working with their farm
machinery (Klein, personal communication, 1997).
While it appears that computer hardware will be more than adequate for preci-
sion agriculture, the same cannot be said for the software. Advances in software
logically lag behind the hardware technology. However, software for precision
agriculture has been more an experience than an application. Berry (1999, in dis-
cussing the human factor in GIS, describes experience as “what you get when you
don’t get what you want.” Computer software in precision agriculture has become
better with time, but precision agriculture is loaded with Berry’s type of experi-
ence. Software will be adequate for precision agriculture when it becomes, as
Berry (1 995) suggests, second nature to the user for assessing information and
translating it into knowledge. For precision agriculture, the knowledge needed is
that for managing variability on the farm, knowledge that is requisite for decision
making. Computers and salient, usable software are going to play a critical role in
the emergence of a precision agriculture system in the near future.

2. Geographic Information Systems

Formally, GIS is an organized collection of computer hardware, software, geo-


graphic data, and personnel designed to efficiently capture, store, update, manip-
ulate, analyze, and display all forms of geographicallyreferenced information [En-
vironmental Systems Research Institute (ESRI), 19971. The GIS concept dates
back to the 1960s when computersbecame available for use in spatial analysis and
quantitative thematic mapping (Burrough, 1986).The science of GIS has evolved
since the 1960s to include data management and modeling, enabling a shift from
mapping to spatial reasoning (Berry, 1993, 1995). The ability to perform spatial
operations on the data distinguishes a true GIS from the many software programs
that do thematic mapping and database management. During the past few years,
mapping software programs have been adding spatial operations, workstation GIS
software programs have spawned microcomputer versions with more limited GIS
capabilities to fit desktop computer technologies, and new microcomputer-based
GIS systems have emerged. There are many different mapping and GIS software
programs that offer different GIS features. None, however, have captured the mar-
ket for application in precision agriculture.
Because precision agriculture is concerned with spatial and temporal variabili-
ty and because it is information based and decision focused (Pierce, 1997a), it is
the spatial analysis capabilitiesof GIS that enable precision agriculture. This state-
ment is true because the value of precision agriculture is derived only when re-
sulting information is turned into a management decision that increases prof-
itability, benefits the environment, or provides some other value to the farm. AGIS,
in the full sense of its formal definition given previously, is key to extracting val-
10 FRANCIS J. PIERCE AND PETER NOWAK

ue from information on variability. Clark and McGucken (1996) refer to GIS as


the brain of a precision farming system. However, available GIS software pack-
ages are complex and difficult to learn for nonspecialists. Some GIS lack data man-
agement and spatial analysis tools needed to understand the variability observed
on the farm and needed to derive site-specific management recommendations.
More functional, easy to use interfaces are needed in order to fully utilize this tech-
nology in production agriculture (Berry, 1995; NRC, 1997).Computer simulation
modeling can help derive the needed understanding of variability (Sadler and Rus-
sell, 1997; Verhagen and Bouma, 1997) and linking GIS to models (Goodchild et
al., 1993) will be important to precision agriculture.

3. Global Positioning Systems

Location control (Schueller, 1992) is essential to precision agriculture for as-


sessing spatial variability and for site-specific application control (Auernhammer,
1994; Tyler et al., 1997). In the early days of precision agriculture, relative posi-
tion within a field was determined by dead reckoning. This was a simple method
in which position was measured relative to a known point determined by measur-
ing distance using radar, ultrasound, and wheel shaft counters. Direction was de-
termined either by using a steering-angle sensor or a gyroscope from the known
point or by direction only if a field had linearized tramlines of known fixed loca-
tion (Auernhammer and Muhr, 1991). Triangulation methods, in which position is
determined relative to two or more known locations using, for example, radio sig-
nals transmitted from reference stations to mobile units (Palmer, 1991, 1995; Scor-
er, 1991), improved position accuracy to as low as 15 cm (95%probability) but
such systems were time-consuming and expensive. By the early 1990s. however,
the GPS known as NAVSTAR (NAVigationSystem with Time And Ranging) was
becoming available for general civilian use including agriculture.This system was
based on 18 satellites that were in orbit by early 1990 (Hoffmann-Wellenhofet al.,
1994; Kaplan, 1996; Kennedy, 1996; Leick, 1995; NAPA, 1995). The United
States NAVSTAR GPS system consists of a constellation of 24 satellites, includ-
ing 3 spares. The first satellite was launched in 1978 but it was not until the Sovi-
et downing of a Korean airliner in 1983 that the decision was made to make GPS
available for civilian use [National Academy of Public Administration (NAPA),
19951. The NAVSTAR GPS system was fully deployed by 1994 and declared
fully operational in 1995. The Russians also deployed a GPS system called
GLONASS (Global Navigation Satellite System) consisting of 24 satellites com-
pleted in 1995. Although there are differences in time standards and coordinate
systems between GLONASS and NAVSTAR, higher end GPS receivers currently
available accommodate the combined use of both GPS systems resulting in in-
creased reliability and accuracy. Although the Russian GLONASS policy called
for ensured availability for 15 years, no charge on a constant global basis, and no
selective availability, the system was degraded to only 14 or 15 active spacecraft
ASPECTS OF PRECISION AGRICULTURE 11

during the fall of 1997 (Perry, 1998). Therefore, changes in GPS technology are
to be expected.
The GPS technology enables precision agriculture because all phases of preci-
sion agriculture require positioning information. GPS is able to provide the posi-
tioning in a practical and efficient manner for a few thousand dollars ( v l e r et al.,
1997). Expensive, high-precision differential GPS (DGPS) systems are available
that achieve centimeter accuracies (Lange, 1996), allow for automated machinery
guidance (O’Conner et al., 1996; 51er et al., 1997) and kinematic mapping of
topography (Clark, 1996), and are useful in the creation of digital elevation mod-
els needed for terrain analysis (Bell et al., 1995; Moore et al., 1993). While the
GPS signal is ubiquitous, there have been problems in making available GPS at
the needed precision for agriculture (Saunders et al., 1996). The U.S.Department
of Defense implemented selective availability (SA) on March 25, 1990, which lim-
ited accuracy of GPS to civilians from about 8-10 m without SA to about 100 m
with SA. This was done by varying the reported precise time of clocks on board
the satellites and by providing incorrect orbital positioning data (NAPA, 1995).
The SA has been overcome by the use of differential corrections transmitted to
GPS receivers (rovers) from GPS receivers at known fixed locations (base). DGPS
involves the transmission of a differential correction, that is, the difference be-
tween actual and predicted position at the base GPS receiver, to rover GPS re-
ceivers, which then apply the corrections to received GPS signals to solve for a
more accurate position (Qler et al., 1997). There are four general ways of pro-
viding a differential correction: a private local GPS base receiver with a radio mo-
dem to transmit to a mobile receiver, a commercial GPS base station at which dif-
ferential corrections are transmitted on FM subcarrier frequencies, a public GPS
base station at which differential corrections are transmitted on AM frequencies
from radio beacons with up to a 250-mile radius [U.S. Coast Guard (USCG) bea-
con system), and a wide area differential GPS (WADGPS) network in which dif-
ferential corrections from a network of base stations are used by the roving GPS
receiver to correct its position (vier et al., 1997). In all cases, DGPS requires ad-
ditional receivers and antennas and is fee based for commercial correction
providers. A differentialcorrection is desirable even without SA because it is need-
ed to achieve the accuracies needed in some aspects of precision agriculture, in-
cluding navigation and guidance. Currently, only WADGPS provides national cov-
erage, whereas all others are dependent on whether the rover is close enough to a
base station to receive the signal consistently. However, this is changing because
FM providers are planning to offer national coverage in the near future and there
are plans for completion of the USCG beacon system nationally (Divis, 1998).
There is currently a debate as to whether the public sector should provide a na-
tional DGPS (NDGPS) to agriculture (NAPA, 1995;Pointon, 1997). Other sectors
of the U.S. economy also need a national NDGPS, so the discussion of who ben-
efits from a publicly supported NDGPS should not be focused on agriculture alone.
The Office of Management and Budget did not support expansion of the USCG
12 FRANCIS J. PIERCE AND PETER NOWAK

radio beacon system for public NDGPS in Ey98. However, some believe that a
government-provided NDGPS system is so important to critical activities that it is
best for the government to provide it (Divis, 1998). Certainly, precision agricul-
ture needs DGPS and will require increased position accuracy as new technolo-
gies for navigation and guidance require higher precision, which may require
DGPS accuracies not available from a government NDGPS. The prophecy of
Auernhammer and Muhr (1991; p. 395) that “their use will also be without costs
in the future” will probably never be realized because DGPS is big business.
Regardless of who provides all aspects of DGPS, farmers and their service
providers need reliable DGPS to achieve the desired positioning for precision
farming operations. Farmers still experience interruptions and interferences in the
GPS and/or differential correction signals, creating gaps in data collection or loss
of application or guidance control. In activities at higher speeds, such as aerial ap-
plications (Kirk and Tom, 1996), time delays in differential corrections may limit
positional accuracy in kinematic mode (NRC, 1997). The specified availability
(four satellites in view at any location) of the NAVSTAR system is 99.85%, with
a reliability (system is in service when it needs to be) of 99.97% (NAPA, 1995).
However, the suitability of the satellite geometry for calculating a solution, re-
ferred to as dilution of precision (DOP), is a problem in farming in which natural
or man-made structures obstruct the receivers’ view of some satellites or interfere
with differential correction reception. There are also geographic locations at which
DOP has been inadequate for needed location precision at certain times during the
day. Additionally,some GPS receivers are susceptible to unwanted interfering sig-
nals from a variety of sources, including farm machinery, making the receiver use-
less in navigation or positioning. Some interferences can be overcome in the de-
sign of the GPS receiver.
Regardless of problems, DGPS has greatly enabled precision agriculture. Of
great importance for precision agriculture, particularly for guidance and for digi-
tal elevation modeling, position accuracies at the centimeter level are possible in
DGPS receivers that use carrier phase in combination with DGPS (Lange, 1996;
Tyler er al., 1997).Accurate guidance and navigation systems will allow for farm-
ing operations not currently in use, including field operations at night when wind
speeds are low and more suitable for spraying and the use of night tillage to re-
duce the light-induced germination of certain weeds (Hartmann and Nezadal,
1990). DGPS technology changes continually and can be followed on the internet
(e.g., Peter Dana’s web site hrtp://wwwhost.cc.utexus.edu/Stp/pub/grg/gcrafr/
notes/gps/gps.html or www.gpsworld.com).

4. Sensors

Sensors are devices that transmit an impulse in response to a physical stimulus


such as heat, light, magnetism, motion, pressure, and sound. With computers to
ASPECTS OF PRECISION AGRICULTURE 13

record the sensor impulse, a GPS to measure position, and a GIS to map and ana-
lyze the sensor data, any sensor output can be mapped at very fine scales. Sensor
technology currently lags behind other enabling technologies (Sudduth et al.,
1997) and the availability of sensors has been cited as the most critical factor pre-
venting the wider implementation of precision agriculture (Stafford, 1996b). Sen-
sors are critical to success in the development of a precision agricultural system
for three important reasons: Sensors have fixed costs, sensors can sample at very
small scales of space and time, and sensors facilitate repeated measures. This
means that the cost per sample is determined by the extent of sensor use, sample
intensity is determined by the capability of the sensor and not the cost or difficul-
ty in sampling associated with traditional physical sampling schemes, and sam-
pling frequency is determined by accessibility of the target and not costs.
The value of sensors and their potential for the future of precision agriculture are
illustrated by yield monitoring.Yield monitoring systems, which use sensorsto mea-
sure crop flow, allow the creation of yield maps with detail not practical with other
measurementtechniques (Pierce er al., 1997).Yield mapping technology may be the
major factor responsible for the growing interest in precision agriculture observed
since its commercial introduction in 1992 (Stafford, 1996b). Prior to 1992, the fo-
cus was on VRT, which would not in itself have sustained precision agriculture.
Yield mapping bolstered precision agriculture and is currently the major precision
agriculture technology in U.S. agriculture. However, the promise of sensing tech-
nologies may make yield mapping technology unnecessary in the future if high-res-
olution remote sensing of the growing crop leads to quantitative prediction of crop
yield prior to harvest. Yield mapping will serve to validate sensor-based predictive
technology, but once operational, yield monitors may not be needed. The use of re-
mote sensing to forecastcrop yields is in use worldwide, and forecastingoffers farm-
ers the ability to market their crops prior to harvest when prices are more favorable.
Sensors are very desirable for use in precision agriculture. Every effort should
be made to promote the application and adaptation of sensors developed in other
industrial sectors, especially the space and defense industries, as well as to pro-
mote the development of new sensor technologies for use in assessing and man-
aging variability in soils, plants, pests, and machinery. Sensors can be contact or
remote, ground based or space based, and direct or indirect. Sensors have been de-
veloped to measure machinery, soil, plants, pests, atmosphericproperties, and wa-
ter by sensing motion, sound, pressure, strain, heat, light, and magnetism and re-
lating these to properties such as reflectance, resistance, absorbance, capacitance,
and conductance. Sensors are needed in precision agriculture because such a sys-
tem requires the collection,coordination, and analysis of massive quantities of data
(Sudduth et al., 1997),some for strategic surveys and inventories and some for use
in real-time applications.
Remote sensing involves the detection and measurement of photons of differ-
ing energies emanating from distant materials. These photons may be identified
14 FRANCIS J. PIERCE AND PETER NOWAK

and categorized by classhype, substance, and spatial distribution, with most de-
signed to monitor reflected radiation (Frazier et al., 1997). Satellite remote sens-
ing dates back to the first aerial photographs taken from balloons in the 1840s.The
first satellite imagery was obtained from TV cameras mounted in satellites in the
early 1960s. Since the U.S.Landsat program launched the first observation satel-
lite in 1972, earth observation has increased and currently India, France, Russia,
Japan, and the European Space Agency also operate earth observation satellites
(Figure 1). Many companies now offer commercial products to agriculture from
images obtained by these satellites or enhanced digital products derived from
them. Remote-sensing satellites collect image data actively by sending a known
signal from the satellite to the earth and measuring the portion of the signal that is
returned. Passive data collection occurs by measuring the incoming energy from
the sun reflected by an object or heat energy emanated from an object. The elec-
tromagnetic energy emanated from an object varies in wavelengths as determined
by the object’s physical and chemical structure. Different images of an object can
be constructed by combining different wavelengths, creating images far more re-
vealing than images obtained from visible light alone. Remote-sensing systems
vary in spatial resolution (meters to kilometers), spectral coverage (portion of the
light spectrum covered), and temporal frequency (days to months). Different ap-
plications in agriculture will require different spatial resolutions, spectral cover-
ages, or temporal frequencies. NASA (1998) provides an online tutorial on remote
sensing and its applications. Moran et al. (1997) provide a comprehensive review
of image based remote sensing for precision agriculture.
Remote sensing holds great promise for precision agriculture because of its po-
tential for monitoring spatial variability over time at high resolution (Hatfield and
Pinter, 1993; Moran e? al., 1997; Stevens, 1993). For example, monitoring of a
growing crop using remote sensing is critical because yield maps document yield
variability but do not provide information on the cause of observed variability.
However, the promise of remote sensing for agriculture has not been realized for
many reasons, including costs, timeliness, and availability (Frazier et al., 1997;
Stafford, 1996b).

5. Application Control

Control is that portion of an automated system in which sensed information is


used to influence the system’s state in order to meet an objective (Stone, 1991).
For precision agriculture, control must be achieved in space and time for varying
single or multiple inputs at different rates, at varying soil depths, and in a uniform
and location-specific manner within fields. Because it is a required component,
control technology has been a strength of precision agriculture since its inception
and the state of application control was recently reviewed by Anderson and Hum-
burg (1 997). Simply stated, if the needed accuracy cannot be achieved at the point
ASPECTS OF PRECISION AGRICULTURE 1s

of application of inputs, then precision agriculture cannot be successful (Ander-


son and Humburg, 1997; Schueller, 1992; Stafford, 1996b). Application control
completes the precision agriculture loop.
Control systems are currently available at varying degrees of precision for vari-
able seed and metering granular fertilizers and pesticides, changing varieties on
the go, anhydrous N application, sprayers, imgation, manure application, and var-
ious tillage implements (Anderson and Humburg, 1997; ASAE, 1991; Robert el
al., 1993, 1995, 1996). The first patented technology for variable rate application
of fertilizers was the Ortlip patent awarded in 1985 to Soil-TEQ, Inc. (now owned
by AgChem), although the earliest references to precision application of fertiliz-
ers appear to be Luellen ( 1985) and Elliot ( 1987).
All issues relating to the accuracy of application equipment are important to pre-
cision agriculture but not all accuracy issues are unique to precision agriculture
(Anderson and Humburg, 1997). General sources of variability in application of
inputs include driving precision, uniformity of distribution, topography, field sur-
face conditions, wind conditions, and metering efficiency. Specific to precision
agriculture are the transition time for changes in rate or product and positioning or
location control and those aspects of application in which changing rates or prod-
ucts affect variability in performance. A high precision, absolute reliable DGPS
will offer the positioning precision required for various tasks in precision agricul-
ture. Some argue for a backup system, such as dead reckoning, to avoid loss of
control if DGPS fails (Auernhammer and Muhr,1991). While very high position
accuracy is available using DGPS, currently the major consideration is cost. Hu-
man driving precision has an expected coefficient of variation of 10% for moder-
ate skill levels (Chaplin et al., 1995) but should be greatly improved with DGPS-
based guidance systems, depending on the accuracy of the DGPS system in use
(51er el al., 1997). O'Conner et al. (1996) report the use of a carrier phase dif-
ferential GPS for automatic vehicle control to achieve a vehicle position accuracy
within a few centimeters and heading to within 0.1".
The issue of transition time is illustrated by the V-shaped spray pattern result-
ing from a transport delay incurred between the injection point and the nozzle dis-
charge for a simple chemical injection system (Steward, 1994, as cited by Ander-
son and Humburg, 1997). Transition times of 3-9 s were reported by Bahri et al.
(1996) when changing seeding rates in grain drills, with transition time depending
on the magnitude of the application rate change. Their data indicated that small
rate changes in seeding rate of 10 kg ha-' did not provide a real rate change, il-
lustrating potential step size rate limitations for some inputs. A transition phase
may limit the spatial resolution of variable rate application of inputs if the target
area is small (Stafford, 1996b)or may cause applicationerrors if the transition time
is greater than the time between detection of the need for change and the equip
ment arrival at the detected position as would occur in real-time application.
Current equipment may not be suitable for precision agriculture. Bashford
16 FRANCIS J. PIERCE AND PETER NOWAK

(1993) and Bashford et al. (1996) report outlet CVs for grain drills ranging from
12 to 22.5%for wheat (Triticumaestivuum L.) and from 16 to 42%for soybean
[Glycine 1 7 1 ~(L). Merr.]. They suggest that external fluted metering devices are
not suitable for precision agriculture. Bahri et al. (1996) measured down-the-row
CVs ranging from 10 to 19%.In general, variation in grain drills has been con-
sidered acceptable if the variability in grain and fertilizer delivery among row units
is below a CV of 15%(Prairie Agricultural Machinery Institute, 1987). This vari-
ance, however, may exceed the desired accuracy in precision seeding systems.
Centrifugal fertilizer spreaders are known to have high sensitivity of the spread
pattern to flow rate variations and efforts are under way to design centrifugal
spreads for precision agriculture (Olieslagers et al., 1996; Kaplan and Chaplin,
1996). Rate changes also affect nozzle performance relative to drop size and flow
rate for the given nozzle design (Anderson and Humburg, 1997) and these issues
are currently being addressed (Giles et al., 1996; Nuspl et al., 1996).
The major issues for precision application of inputs remain transition time for
changes in product or rate, uniformity of application, and rate increment control.
There are other issues affecting the availability or performance of application con-
trol, including the development of standards for communication and connectivity
among manufacturers, a topic being addressed by many organizations including
an association of industry and the public sector called the Ag Electronics Associ-
ation, ASAE, and the International Standards Organization. There are laws regu-
lating fertilizer quality that in some states (e.g., Arkansas and Michigan) limit the
blending of fertilizers on the go because of the need for a guaranteed chemical
analysis, and such laws will have to be properly addressed. There are issues relat-
ed to equipment wear (Ballal et al., 1996) and to weather conditions at the time of
application. Kirk and Tom (1996) report that up to 13%of the variability in their
tests for spray aircraft was due to wind conditions. Heterogeneity in the composi-
tion of some materials affects the flow or spreading properties (e.g., manures; Ess
et al., 1996).Topography and field surface conditions also affect accuracy, in part
due to their effect on flow of materials in the hoppers or tanks.
Application control, including navigation and guidance, has been enabling pre-
cision agriculture since its inception and continues to improve. Farmers and their
service providers have the capability to apply very precise applications of inputs
site specifically. Application control technology will continue to improve and sup-
port the needs of precision agriculture. What is needed is knowledge of what in-
puts are required where and when.

B. STEPSINPRECISIONAGRICULTURE
The basic steps in precision agriculture are assessing variation, managing vari-
ation, and evaluation. While the enabling technologies facilitate precision agri-
ASPECTS OF PRECISION AGRICULTURE 17

culture, it is the knowledge and understanding of variability and the extent that
site-specific agronomic recommendations are available to manage this variability
that make precision agriculture viable. Also, because nothing is known with cer-
tainty and many factors affect crop production, evaluation must be an integral part
of any precision agriculture system. These are interdependent and necessary com-
ponents of a precision agriculture system yet to emerge. As will be seen, there is
significant variation in the progress and research underlying each of these com-
ponents. Assessing variability, precision management, and evaluation are the fo-
cus of the remainder of this chapter.

1. Assessing Variability

Assessing variability is the critical first step in precision agriculture since it is


clear that one cannot manage what one does not know. The processes and proper-
ties that regulate crop performance and yield vary in space and time. Adequately
quantifyingthe variability of these processes and properties and determining when
and where different combinations are responsible for the spatial and temporal
variation in crop yield is the challenge facing precision agriculture (Mulla and
Schepers, 1997). Techniques for assessing spatial variability are readily available
(Beckettand Webster, 1971; Cressie, 1991; Goovaerts, 1997,1999; Isaaks and Sri-
vastava, 1989; Mausbach and Wilding, 1991; Mulla, 1997; Mulla et al., 1990;
Rossi et al., 1992; Trangmar et al., 1985; Warrick et al., 1986; Webster and Oliv-
er, 1990; Wollenhaupt et al., 1997) and have been applied extensively in precision
agriculture.The bulk of the literature on precision management relates to some as-
pect of assessing spatial variability. Techniques for assessing temporal variation
also exist (Shumway, 1988) but the simultaneous reporting on spatial and tempo-
ral variation is rare and the theory of these types of processes is still in its infancy
(McBratney et al., 1997). Such space-time statistical applications are important
to precision agriculture because many phenomena exhibit spatial patterns that de-
velop over the course of time. They are also important because a cause-effect re-
lationship may exist in time but not in space (Stein et al., 1997). A good example
of the latter point is a yield map. A yield map defines the spatial distribution of

crop yield but does not explain the observed variability. Imagery of crop growth
and development over the growing season can uncover the cause-effect relation-
ship that explains not only the yield variation within a field but also the magnitude
of yield observed in a particular growing environment (Schepers et al., 1996). As
we will discuss later, some variables exhibit strong spatial dependence but low
temporal dependence (high temporal correlation), making them more conducive
to current forms of precision management. To repeat our premise, structured spa-
tial dependence is needed for precision management, whereas temporal variabili-
ty increases the difficulty with which it can be implemented. This will become
more clear in our discussion of managing variability.
18 FRANCIS J. PIERCE AND PETER NOWAK

For precision agricultureto be useful, variation must be known, of sufficient mag-


nitude, spatially structured (nonrandom), and manageable (Pierce, 1995). Know-
ing variation implies a measure of accuracy, either in measurement or in predic-
tion. An accurate assessment of variability is essential, but the prolific use of maps
without measures of accuracy indicates that this important aspect is often ne-
glected. Knowledge also implies a sense of understanding. It is not uncommon to
have detailed measures of variation within a field with little understanding of the
causes of the observed variability, as suggested in our yield map example. The
magnitude of both the mean and the variation determines the potential for benefits
from precision agriculture.For a given parameter, a threshold variance must be ex-
ceeded for precision management to be an improvement over whole-field man-
agement for crop production.Also, the absolute magnitude of a parameter must be
in a manageable range. For example, field soil fertility is already in the high range
and no fertilizer would be recommended anywhere. Thus, a parameter can vary
spatially but not deviate sufficiently from the mean field value or values do not fall
within a manageable range to justify precision management. A high degree of spa-
tial dependence is needed for current applications of precision agriculture. Spatial
dependence drives precision agriculture because parameters with high random
variation (low spatial dependence) will not be conducive to site-specific manage-
ment and will be best managed on the average. Finally, accurate knowledge of
large, spatially dependent variation is not sufficient for precision agriculture if the
variation is not manageable. Scale is very important in this regard. For example,
even with accurate maps, nutrient deficiencies or pests may be difficult to manage
precisely if their areal expression is considerably smaller than the minimum area
treatable by available application equipment. Biological processes that vary on
very small scales are difficult if not impossible to exploit with current precision
management (Groffman, 1997). Drainage ways within fields may be difficult to
manage because they vary from high to low yields across years depending on sea-
sonal precipitation patterns. While variation in drainage is known, spatially struc-
tured, and of sufficient magnitude, how to precisely manage time-dependent enti-
ties such as drainage is uncertain. Management decisions in these situations are
more likely to be driven by risk assessment strategies.
Ultimately, farmers must be able to delineate areas that will respond similarly
to inputs that optimize crop performance (yield, quality, and environment). Maps
form one basis for precision management; real-time management forms the other
basis. Use of management maps is more common and these can be categorized as
condition maps, prescription maps, and performance maps. In real-time precision
management, maps are not necessary because inputs are triggered by real-time
measurements of soil, crop, or pest condition. Condition maps are measured and/
or predicted using a broad array of technologies and techniques for estimating the
spatial distribution of one or more properties or processes. Measurementsobtained
in real-time precision management can be mapped as condition maps for later use.
ASPECTS OF PRECISION AGRICULTURE 19

Prescription maps are derived from one or more condition maps and form the ba-
sis for VRT (Sawyer, 1994). Performance maps record either inputs (fertilizers,
pesticides, seeds, energy, etc.) or outputs (crop yield and quality) and include de-
rivatives of performance maps, such as profit maps (outputs-inputs). Performance
maps are possible for any part of a farming enterprise that can be sensed in real
time and a location recorded. Performance maps can ultimately serve as condition
maps. For example, multiple-year yield maps can be used to estimate yield goal
maps often needed for precision prescriptions or to estimate soil test levels using
mass balance approaches.
Condition maps are a critical component of precision agriculture and can be gen-
erated in four major ways: (i) surveys, (ii) interpolation of a network of point Sam-
ples, (iii) high-resolution sensing, and (iv) modeling to estimate spatial patterns.
All these methods are scale dependent and have limitations to their use in preci-
sion management. We briefly discuss these here but refer the reader to more de-
tailed references.

a. Surveys
Surveys are purposeful inventories of specific quantities and have been partic-
ularly useful in natural resource management. Surveys are designed with specific
purposes in mind and, in general, have limitations when used for other purposes
or intense applications such as precision agriculture. The National Cooperative
Soil Survey of the USDA-NRCS is an extensive inventory of soil resources that
includes soil data and maps needed for crop production (Soil Survey Division
Staff, 1993). Surveys of varying age and scale of measurement are available for
most of the United States. At first glance, the soil survey should be an important
asset to the principles of precision agriculture. Farming by soils was initially
thought to be a reasonable basis for precision management (Cam et al., 1991; Lar-
son and Robert, 1991).However, existing soil surveys have proved of limited val-
ue in explaining spatial variability observed within fields. Mausbach et al. (1993)
state that “Kellogg (1961; p. 58) was very clear that the soil survey and its inter-
pretations are not site specific.” They concluded that while soil surveys are useful
for planning on-farm resource management systems and for highway, urban, and
other planning activities, they are not designed for specific applications such as
soil-specific farming.
Are soil surveys important to precision agriculture? We should not expect that
variation in crop performance, either agronomic or environmental, will be ex-
plained by soil and landscape properties alone. The crop production process ia
more complex than this deterministic assertion would suggest. However, within a
given climate regime, we should expect crop performance to generally correspond
to differences among soils and landscapes. Yield mapping has supported this gen-
eral relationship.The value of the soil survey to precision agriculture could be im-
proved by intensifying map scales to fine scale resolutions needed in detailed en-
20 FRANCISJ. PIERCE AND PETER NOWAK

vironmental modeling applications or site-specific management (Moore et al.,


1993). Accomplishing this is no small task. Nonetheless, new data sources (e.g.,
digital ortho photos, airborne and satellite imagery, and yield maps) and analysis
techniques (e.g., terrain analysis) make it possible to map soils at needed resolu-
tions (Bell et al., 1995; Moore ef al., 1993). Mausbach el al. (1993) suggest that
such data collection activities should be a private-sector activity and not performed
by the NRCS. However, there are two areas in which a private and public-sector
cooperative relation could be developed. First, the NCSS program should com-
plete the current soil survey in a manner that embraces available technologies for
collecting data, generating finer resolution soil surveys, and publishing surveys in
digital formats. Second, as will be noted later, the actual implementation of preci-
sion agriculture is spatially “lumpy” or unequal in its distribution across agricul-
tural areas. Not all agricultural areas are of equal agronomic importance to the
United States. Certain critical areas from either an agronomic or an environmen-
tal perspective could be designated for enhanced NCSS activities. The soil survey
can be important to precision agricultureif the same logic (i.e., what is needed and
where it is needed) is applied to defining public and private-sector roles in the al-
location of scarce fiscal resources.

b. Interpolation of Point Samples


Another technique for assessing spatial variability involves sampling process-
es. A network of points in some spatial arrangement is sampled and then interpo-
lated to produce a spatial estimation (usually a map) of the whole area using a range
of statisticalprocedures. Readers are referred to references previously cited for de-
tails on the use of these procedures. These spatial statistical techniques can also be
repeated over time to estimate the temporal variability (McBratney ef al., 1997;
Stein et al.. 1997). Network or spatial sampling of points and interpolation into
maps is useful when it is feasible to directly measure only a small number of points
due to economic or temporal constraints. To a large extent, sampling depends on
the nature of the entity of interest. Soil sampling for soil survey, for example, is
used to determine how much of the land is of a particular type or what proportion
possesses some soil attribute (Webster and Oliver, 1990). For pest management,
interest may be in obtaining insect pest density maps either within a field or with-
in a region and over time (Fleischer er al., 1997). Regardless of entity, the nature
of spatial and temporal variation of that entity should affect sampling and statisti-
cal estimation procedures. The goal of network or spatial sampling for precision
agriculture is to provide an accurate and affordable map of the occurrence of a spe-
cific parameter to be managed. What this parameter is will depend on the nature
of the cropping system and its biophysical context. Three important issues need to
be addressed regarding spatial sampling for assessing variability: sample unit,
sample design (arrangement and intensity), and map accuracy.
The first important question to address is what must be included in the entity un-
ASPECTS OF PRECISION AGRICULTURE 21

der investigation and how it is defined operationally-the sample unit definition. If


soil is the entity of interest, then it is important to define what constitutes a soil, for
example, whether a soil is defined in terms of a map unit, a pedon, or a sample depth
and if the sample is cornposited (Webster and Oliver, 1990; Wollenhaupt et al.,
1997). For insects, Fleischer et al. (1997) cite Pedigo’s (1994) definition of sam-
pling unit as the proportion of habitable space from which insect counts are taken.
Here, sample unit includes unit areas, plant parts such as leaf surfaces,and time trap
ping (Fleischer et al., 1997).There is little standardizationof current applicationsof
precision agriculture in defining the sample unit when engaging in spatial sampling.
Sample design refers to the spatial and temporal arrangement of samples and
the number of samples needed to accurately estimate the spatial distribution of the
parameter of interest. Sample designs will vary for different parameters and for
different biophysical environmentsas parameters vary at different spatial and tem-
poral scales and may vary in different environments.There must be a link between
the capability of the enabling technologies with the underlying agronomic rules
for precision management to the level of precision (scale) needed in variation as-
sessment. Little progress in precision management will be made if something
varies at the submeter level but can only be managed at scales compatible with
large field application equipment (Groffman, 1997). The scale of spatial and tem-
poral assessment depends on the spatial and temporal heterogeneity in the bio-
physical environment and cropping system to be managed. For example, some
fields contain dissimilar soil types, whereas others are more homogeneous. They
need not be sampled with the same design, nor do they have the same potential for
precision management. Insects vary spatially but may require different sampling
designs for each progressive generation during a growing season (Fleischer et al.,
1997).Again, there has been little discussion in the precision agriculture literature
of the congruency between the sample design with the enabling technologies,
agronomic principles, and biophysical diversity.
The objective of a specified sample unit and congruent sample design is to pro-
duce a quality map that has value for management decisions. There is no guaran-
tee that a given sampling scheme will produce an accurate map or that an interpo-
lation method is optimal (Gotway et al., 1996b). Techniques to evaluate map
quality include cross-validationand mean square estimates from the regression of
estimated versus measured values obtained from validation sets. Management de-
cisions based on inaccurate maps appear to increase variability rather than man-
age it. More consideration needs to be given to map accuracy and its implication
for precision management.

c. High-Resolution Sensing
The importance of sensing technology to precision agriculture was discussed
earlier. The improvement of high-resolution sensing over interpolation of sam-
pling points is evident when yield mapping is considered. For example, sampling
22 FRANCIS J. PIERCE AM) PETER NOWAK

intensity in a yield map is orders of magnitude higher than that used in grid sam-
pling. The exercise of grid sampling a yield map would reveal the weaknesses in-
herent in interpolation of spatial sample points. Yield maps do not indicate the
causes of the yield magnitude or its variability. However, it is the causes of vari-
ability that need to be quantified if farmers are to adjust their management prac-
tices to specific conditions within a field at appropriate times during the growing
season. High-resolution remote sensing of the growing crop will reveal stresses
that impact the crop during the growing season (Scheper et al., 1996).Addition-
ally, it is not physically or economically possible to accurately map certain soil
properties, crop condition, or pest status without the use of high-resolution sens-
ing. The lower cost and ease of measure of high-resolution sensors will be critical
to the future success of precision agriculture.

d. Modeling
Modeling is proposed as an important tool in precision agriculture to stimulate
spatial and temporal variation in soil properties (Verhagen and Bouma, 1997),
pests (Kropff et aL, 1997), crop yield (Bamett et al., 1997; Sadler and Russell,
1997). and environmental performance of cropping systems (Verhagen et al.,
1995a).Models have been developed and calibrated for specific purposes but have
not been used extensively in spatial prediction. A major problem of models is the
availability of inputs needed to run them. A major advantage of models is their
ability, once calibrated, to simulate the temporal component of crop production.
This capability should allow models designed to account for spatial variability to
evaluate different precision management scenarios that would otherwise be pro-
hibited by time and cost considerations. The application of models to the simula-
tion of the space-time continuum of crop production is a critical research need
(Sadler and Russell, 1997).

2. ManagingVariability

Once variation is adequatelyassessed, farmers must match agronomic inputs to


known conditions employing management recommendationsthat are site specific
and use accurate application control equipment. As already discussed, the poten-
tial for accurate application control on the farm exists but the extent to which it
has reached the farm is limited. High-precision application control technology is
available in many areas commercially,but it is not generally installed on farmer-
owned equipment. Application technology is not the only factor that can limit pre-
cision management, however. Our discussion now focuses on the agronomic fea-
sibility of precision agriculture, i.e., whether precision agriculture improves crop
performance. Precision agriculture has operated for some time on the assumption
that best managementpractices developedfrom decades of agronomicresearch are
applicable at any scale of management. This assumption can and will be tested as
ASPECTS OF PRECISION AGRICULTURE 23

precision agriculture evolves on the farm. What follows is a discussion of the state
of precision agriculture relative to precision management of inputs.

a. Precision Soil Fertility Management


Nutrient input to crop production is important because soils naturally do not sup-
ply nutrients in sufficient quantities to meet nutrient demands of commercial crops.
Approximately 2 1.3 million tons of commercial N, P,O,, 60(1 1.7,4.4, and 5.1
million tons, respectively)fertilizers were applied and 1.23, 1.32, and 1.44 million
tons respectively of economically recoverable nutrients in manure were available
for application in the United States in 1995 [USDA-Economic Research Service
(ERS), 19971. While important inputs to crop production, fertilizers and manures
are also identified as major sources of nutrient contamination of surface and
groundwater in agricultural areas of the United States (Mueller etal., 1995). Soils
vary in their ability to supply nutrients to plants, and crops vary in their demand
for nutrients. The fact that soil supply and plant demand vary in space and time
and nutrient losses through leaching, erosion, and runoff also vary temporally and
spatially (Sharpley, 1997)indicates that significantopportunitiesmay exist for pre-
cision management of soil fertility. In support of precision nutrient management
is the fact that soil testing has been the basis for fertilizer recommendations since
the late 1940s and recently has been important in determining where nutrients
should not be applied as well as where they should be applied (Hergert etal., 1997).
The potential for improved precision in soil fertility management combined with
increased precision in application control (Anderson and Humburg, 1997) make
precision soil fertility management an attractive, but largely unproven, alternative
to uniform field management (Sawyer, 1994).
For successful implementation, the concept of precision soil fertility manage-
ment requires that within-field variability exists and is accurately identified and re-
liably interpreted (fertilizer recommendations are site specific), that the variabili-
ty influences crop yield, crop quality, and/or the environment, and that inputs can
be applied accurately (Sawyer, 1994; Pierce, 1995). We hypothesize that the ease
with which precision management is accomplishedand its value vary with specific
nutrients or lime. The higher the spatial dependence of a manageable soil proper-
ty, the higher the potential for precision management and the greater its potential
value. The degree of difficulty, however, increases as the temporal component of
spatial variability increases. Applying this hypothesis to soil fertility would sug-
gest that liming and P and K fertility are very conducive to precision management
because temporal variability is low. For N, the temporal component of variability
can be larger than its spatial component (Pan et al., 1997), making precision N
management much more difficult in some cases. We will discuss precision man-
agement for lime, P and K, and N separately. While there is little work published
on precision management of micronutrients,we expect that concepts discussed for
lime and the macronutrients should apply.
24 FRANCIS J. PIERCE AND PETER NOWAK

i. LIME. The value of liming acid soils is well known (Adams, 1984)and tech-
niques to measure lime requirement are well established (McLean, 1982). Over-
application of lime is costly and may have detrimental effects on factors affecting
crop yield, particularly nutrient availability (McLean and Brown, 1984). Spatial
variation in soil pH is to be expected (Linsley and Bauer, 1929; Peck and Mel-
sted, 1973). Thus, the importance of pH management in soils through liming and
the expectation that pH varies spatially make the prospects of variable liming good
where soil acidification is expected. Acidity can have considerable spatial depen-
dence but the spatial dependence has a low temporal dependence (Franzen and
Peck, 1995; Hergert et al., 1997). That is, exchangeable acidity varies with time
but areas of acidity will remain acid until limed. This fact makes variable lime
management relatively easy once the spatial variability is accurately established.
Through grid soil sampling, the spatial dependence of soil pH has been verified
(Peck and Melsted, 1973; Laslett et al., 1987; Tevis et al., 1991). Peck and Mel-
sted (1973) sampled soils from two 16.2-ha fields in Illinois in 1961 on a system-
atic grid spacing of 25.2 m and found that pH averaged 6.55 and 6.21 for the two
fields but ranged from 5.5 to 8.0, with little correlation to soil map unit. Over time,
the Mansfield field was sampled and limed periodically, but the spatial pattern of
soil pH remained similar between 1961 and 1991 (Franzen and Peck, 1995; Her-
gert et d., 1997). Pierce et al. (1995) reported that soil pH ranged from 2.0 to 3.1
pH units in three fields in Michigan, whereas mean pH values ranged from 6.0 to
6.7, with a strong spatial dependence at each site. There are, however, surprising-
ly few published studies that evaluate variable liming even though this practice is
often identified as a major benefit of precision agriculture. Borgelt et al. (1994)
showed that for an 8.8-ha field, which on average required a 3.4- to 4.5-mg ha-'
lime application, 9-12% of the field would have been overlimed by a uniform ap-
plication of lime and 37-41% of the field would have been underlimed, with dif-
ferent lime application rates depending on the method of lime determination used.
Crop response was not evaluated in this study since variable lime applicationswere
not made. Also, they did not evaluate map accuracy, although they used a modi-
fied composite design sampling pattern (sampling density of approximately 7.7
samples per acre) that gave a diversity of sampling points at varying distances to
more precisely measure the semivariogram.Franzen and Peck (1995) applied lime
to the Mansfield field originally sampled by Peck and Melsted (1973) at a uniform
rate of 4.48 mg ha-' and evaluated the change in soil pH and Ca and Mg concen-
trations in leaf tissue of corn (&a mays L.) and soybean. Hergert el al. (1997) re-
ported that grain yields were positively correlated to soil pH in 1991 before lim-
ing but soybean yields were unrelated to soil pH in 1992 after lime application.
What is the potential for variable lime in the United States? It must first be de-
termined whether variable liming will pay. The expectation is that soils requiring
lime will benefit from increased yields and that overliming is costly and poten-
tially yield reducing. The field in the Borgelt et al. (1994) study on average need-
ASPECTS OF PRECISION AGRICULTURE 25

ed lime so it should be determined whether the more precise lime application im-
proved the benefits over a uniform application. From this study, it is not possible
to tell if the average application might have been sufficient to alleviate plant-lim-
iting effects of low pH. About 12% of the Borgelt et al. (1994) field (1.1 ha) re-
quired no lime and not applying lime would have resulted in a cost saving. What
effect overliming would have on crop productivity is uncertain. Another issue is
missing liming opportunities by average field testing. In the fields evaluated by
Pierce ef al. (1999, the average pH was adequate for crop growth and no lime
would have been applied to any of the three fields. In such cases, only underlim-
ing effects would occur, assuming the lime application map was correct in the first
place. Very rarely is the quality or accuracy of the map assessed and its importance
can be easily demonstrated (Laslett et al., 1987). Figure 3 presents three maps of
soil pH for a field in central Michigan generated from three grid sample spacings
(F. Pierce, unpublished data). The average pH for this field is 6.5. Note that the
area needing lime varies considerably with grid scale. In fact, grid intensity and
sampling design as well as the map interpolation method all impact map accura-
cy (Gotway er al., 1996b). Figure 3 clearly demonstrates that the amount of over-
limed or underlimed area varies with grid scale. The field average pH, however,
results in considerable underliming in this field. Because lime requirement can
have high spatial dependence but low temporal variability, spending the time and

Figure 3 Lime recommendation maps interpolated using inverse distance squared for 30-, 61-,
and 91-m grid soil samples obtained from a central Michigan field.
26 FRANCISJ. PIERCE AND PETER NOWAK

resources to obtain quality lime application maps makes sense. This notion is fur-
ther supported by the possibility that soil property, fertilizer application, and crop
yield maps could be used to predict the temporal variation of soil pH, making ex-
tensive future sampling unnecessary once the spatial variability is known (Her-
gert et d.,1997). Real-time sensing of soil pH could also refine our ability to make
accurate and cost-effective lime requirement maps and, over time, predict the need
for additional variable lime applications.
The caveats to variable liming are clear. For variable liming to be profitable, in-
creased yields or lime application savings are needed to compensate for the cost
of variable liming. However, acid soils in need of lime do not necessarily reduce
crop yields because grain crops may not necessarily be affected by acid soil con-
ditions (Black, 1993). McLean and Brown's (1984) summary of crop response to
soil pH in the Midwest showed that corn frequently did not respond to soil pH of
5 or 6, whereas alfalfa was strongly affected by this pH range, with soybean in-
termediate in response. It was the beneficial effects of lime on legumes that formed
the basis for lime applications prior to the 1950s, after which the need for lime was
based on neutralizing the soil acidity resulting from additions of large quantities
of residually acid fertilizers (McLean and Brown, 1984). Farmers may or may not
experience yield changes due to liming. This may also be the case for overliming
because liming soils of high pH may or may not have detrimental effects on crop
yield. Negative effects of overliming are usually tied to decreased nutrient avail-
ability at high pH (Adams, 1984). Christensen et al. (1998), however, reported that
applications of sugar beet lime to the high pH lakebed soils of the thumb region
of Michigan increased soil pH by 0.3-0.5pH units but had no detrimental effects
on crop yield and improved sucrose content in the first 2 years of the 5-year study,
Where will variable liming work? The potential for variable liming is related in
part to how much lime is applied and where it is applied. Lime was applied to <6%
of acreage of any major crop in the United States in 1995 with an average appli-
cation rate of < 2 Mg ha-' (USDA-ERS, 1997). The amount of lime applied could
increase with variable liming if there are sufficient fields with an average pH near
optimum but with a broad range in pH such as the fields evaluated by Franzen and
Peck (1995) and Pierce et al. (1995). To some extent, the need for variable lime
can be anticipated by identifying where soil acidification is a recumng problem
in production agriculture. There may be equipment limitations to variable lime
application because agricultural lime is difficult to apply uniformly with con-
ventional spreaders. The availability of appropriate precision equipment may be
limited due to capital equipment costs and lack of sufficient acreage for lime ap-
plication services to pay for such equipment. Variable lime applications are often
made by applying uniform rates to areas within fields delineated as needing lime,
but the extent to which this is occurring is difficult to assess.
The potential for variable lime exists where soil acidification is recurring. Ac-
curate lime application maps are needed to ensure precision applications and avoid
ASPECTS OF PRECISION AGRICULTURE 27

over- and underliming. Expectations for crop improvement from liming should be
lowered to reflect the lack of response to low pH by some crops at least in some
years. In summary, the character of high spatial variability with a low temporal
component for soil pH makes it ideal for precision agriculture if this management
decision can be shown to be profitable.

ii. PHOSPHORUS AND POTASSIUM.Precision management of P and K was an


early focus of precision agriculture because there was an established basis for fer-
tilizer recommendations in soil testing that could theoretically be applied at any
scale. Moreover, the technology to variably apply fertilizers became available in
the mid-1980s. Spatial variability in P and K was already known (Peck and Mel-
sted, 1973) and was not difficult to measure within agricultural fields. The concept
of VRT was very intuitive and easy to understand and implement (Luellen, 1985).
The temporal component of the spatial variability of P and K is low, making it easy
to soil test at a convenient time and requiring only periodic (every few years is of-
ten the recommendation) repeated sampling. Further support for precision man-
agement derives from the fact that P soil tests are increasing (Hergert er al., 1997)
and P in runoff and sediment is an increasing environmental problem (Sharpley,
1997), thereby increasing the importance of soil testing and matching P inputs with
crop need.
The current basis for precision management of P and K, therefore, is fertilizer
recommendations based on traditional soil fertility tests using various sampling
schemes to assess within-field variability (Hergert et al., 1997). There are some in-
teresting issues in this regard. While traditional soil testing and fertilizer recom-
mendations are currently used, there are questions as to whether these are appro-
priate for site-specific management (Hergert er al., 1997). The improvement or
verification of soil test and interpretations is considered by Fixen (1998) to be a
major research need for site-specific nutrient management. Interestingly, there is
a countertrend in fertilizer recommendations toward increasing the scale of soil
test recommendations rather than making them more site specific. Vitosh er al.
(1 996), for example, aggregated soil test recommendations from Indiana, Michi-
gan, and Ohio into a single tristate set of recommendations. We suspect a similar
change in scale of underlying agronomic recommendations is occurring elsewhere
in the United States. Another complicating issue is that for a given soil test result,
there are many different rate recommendations available to a farmer depending on
the source of the recommendation (Cox, 1994; Olson er al., 1987). There are no
standards for soil test sampling designs, sampling intensity, or methods of inter-
polation used in creating nutrient management maps, although some offer sug-
gestions (Gotway et al., 1996a,b; McBratney et al., 1996; Wollenhaupt er al.,
1997). In short, there appear to be no standards regarding the underlying agro-
nomic principles that should be guiding the development and application of pre-
cision agriculture.
28 FRANCIS J. PIERCE AND PETER NOWAK

Commercially, grid sampling is frequently done using a grid sampling intensi-


ty of one sample per hectare or less. Since the cost of grid soil sampling ($ ha-')
is inversely proportional to the grid spacing squared, it is very easy for sampling
costs to exceed the value in fertilizer savings as sampling intensity increases.There
is no guarantee that a specific grid sampling intensity will result in an accurate nu-
trient or fertilizer recommendation map (Birrell et af., 1996). It is rare to find ver-
ification of the accuracy of nutrient or fertilizer recommendation maps either in
practice (farmers)or in the literature. Long er af. (1999, however, provided an ex-
ception to this generalization because they evaluated the quality of management
maps by comparing simple measures of accuracy and precision. The need for ac-
curate identification and reliable interpretation of within-field variation was con-
cluded by Sawyer (1994) to be the basis for VRT and is prerequisite for its suc-
cess. Also, it is also not uncommon to find little correlation between grid P or K
values and crop yield (Pierce et af., 1995; Smith er af., 1998). It is clear that more
complex analyzes are needed to assess these relationships (Mallarino er af.,1996).
Lowenberg-DeBoer and Swinton (1997) reported inconsistent results and gener-
ally low profitability of precision P and K management for the studies they re-
viewed, suggesting that management of one or two nutrients will not form the
basis for profitability of precision agriculture. These are current or potential diffi-
culties with precision management of nutrients such as P and K. However, the low
temporal component of variability for these nutrients suggests that precision man-
agement will be of increasing value as spatial dependence increases.
Studies assessing the management of P and K nationwide are being performed,
although the literature is lagging behind the activity. Also, not all reports are new.
Peck and Melsted (1973) reported that P soil tests for two fields in Illinois sam-
pled on a 25.2-111 systematic grid in 1961 had means of 32.9 and 30.7 kg ha-' and
ranges of 10-165 and 10-100 kg ha-', with CVs of 76 and 49%. Pierce et af.
( 1995) reported that spatial variability of P and K for three fields in Michigan sam-
pled on a 30.5-m regular grid had means (mg kg-I) and CVs (%) of 23 (34%), 50
(50%),and 124mgkg-' (26%)forPand210(39%), 173 (50%),and 121 mgkg-'
(297%) for K with strong spatial dependence for each nutrient. Han et af. (1996)
reported spatial variability of P and K for two adjacent center pivots in Washing-
ton on a 6 1-mregular grid that had means (mg kg- *) and CVs (%) of 24 mg kg- '
(24%) for P and 183 mg kg-' (21%) for K but very weak spatial structure even
after data detrending.Walker ef af. (1996) examined the effects of 80 years of cul-
tivation and slope position on soil test K levels. For the crest, midslope, and de-
pression slope positions, the mean (mg kg-') and CVs (%) for soil test K were
'
283 (23%), 450 (41%), and 753 mg kg- (16%) for cultivated sites and 625 (18%),
509 (22%), and 695 mg kg-' (21%) for native sites in an east-central Alberta,
Canada, location. Mallarino er al. (1996) reported variability of soil tests for three
Iowa fields of 3-6 ha in size that were sampled on 15-m grid spacing. Soil test
means (mg kg-I) and CVs (%) were 88 (13%), 20 (25%), and 45 mg kg-' (16%)
ASPECTS OF PRECISION AGRICULTURE 29

for P and 243 (1 3%), 107 (1 7%), and 2 13 mg kg- (28%) for K. Response to vari-
able P and K management has been mixed (Lowenberg-DeBoer and Swinton,
1997). Carr et al. (1991) and Wibawa et al. (1993) did not find profitable returns
for variable management, whereas Mulla et al. (1992) and Wollenhaupt et al.
(1994) did.
When and where will precision P and K management be profitable?The answer
may be found in soil testing philosophies that form the basis for fertilizer recom-
mendations. There are two fundamental fertility management philosophies-suf-
ficiency versus buildup maintenance (Dahnke and Olson, 1990). The buildup
maintenance concept promotes the application of sufficiently high rates of P and
K to raise soil test levels in 1 or 2 years, followed thereafter by an annual appli-
cation equivalent to the amount to be removed by the crop to be grown. This is a
preventative approach to protect against yield loss because of nutrient deficiency
and focuses on building soil fertility. Once the initial buildup is achieved, then the
need for further soil testing is eliminated if fertilizer application adheres to the
maintenance application (Dahnke and Olson, 1990).Applying this soil testing and
fertilizer management philosophy to precision management, the goal would be to
either increase the soil test levels of a field through fertilization or decrease the soil
test levels through crop removal to the maintenance level and then fertilize based
on crop removal as indicated by yield maps. Where nutrients are high within a
field, no fertilizers would be applied until soil test levels were depleted to the main-
tenance levels, and where nutrients are low fertilizers would be applied based on
a buildup program. The long-term impact of this management philosophy is to ho-
mogenize soil test levels and then rely on yield maps for fertilizer recommenda-
tions, thereby eliminating the need for grid soil sampling (Goedeken et al., 1998).
The sufficiency approach, preferred by universities, is similar to the rapid buildup
and maintenance except the rate of buildup is much slower and the buildup level
is much lower (Dahnke and Olson, 1990). Fertilizers are applied based on proba-
bilities of yield response to applied fertilizer at soil test classes of low, medium,
and high. No fertilizers are recommended for high testing soils, a maintenance lev-
el is recommended for medium testing soils, and a larger amount than the mainte-
nance level is applied on low testing soils. The sufficiency approach is crop spe-
cific, whereas the buildup maintenance approach builds soil fertility. Because the
sufficiency approach is more conservative, less fertilizer will be recommended.
Applying this approach to precision management, soil testing is still required be-
cause soil test category is required to make fertilizer recommendations.Ultimate-
ly, however, the sufficiency approach will approach the maintenance plateau as the
fertility of low testing soils increases and that of high testing soils decreases over
time. It appears that the only differencebetween the two management philosophies
is how much and how long it will take to reach the maintenance plateau. To some
extent this is related to the critical level of soil tests proposed by the different man-
agement philosophies.
30 FRANCIS J. PIERCE AND PETER NOWAK

Ultimately, if precision nutrient management is based on soil testing and fertil-


izer management philosophies, over time variable rate applications should create
soil test levels that are optimal throughout the field and would require future vari-
able rate applicationsbased on crop removal and nutrient fixation in soils for main-
tenance. One might call this precision agriculture on the maintenance plateau.
Goedeken et al. (1998) suggest that over time variable rate P fertilization should
reduce the need for variable rate applications because soil tests will become spa-
tially uniform. However, even if spatially uniform soil tests resulted from variable
rate applications, unless crop nutrient removal was uniform there would always be
a need to variably apply nutrients because the fertilizer management philosophy
of maintenance requires fertilizers to match crop removal. If accurate nutrient
maps were obtained for a field, then spatial mass balance calculations should pre-
clude the need for further soil testing, except for intermittent monitoring of con-
trol points to check predictions. This also assumes that none of the nutrients are
being lost to the larger environment (discussed later).
The key to precision management of P and K is first and foremost to obtain ac-
curate maps. Sawyer (1994) made this perfectly clear in his review of precision
nutrient management. Accurate fertility maps require appropriate sampling de-
signs and adequate sampling densities (Gotway er al., 1996b; Laslett er al., 1987;
Wollenhaupter al., 1997) in addition to well-calibrated soil test procedures and as-
sociated interpretations (Hergert er al., 1997). Rarely is the accuracy of P and K
maps evaluated. For one Michigan field, interpolation of a 91-m grid predicted
only 3 1% of the variation in soil test P levels obtained from a 30.5-m independent
sampling grid, whereas the 30.5-m grid interpolation predicted 73% of the varia-
tion in measured soil test P levels at the 91-m grid (Pierce, unpublished data).
While the increased precision was evident from the improved regression, a sig-
nificant deviation of the regression line from the 1:1 line indicated considerable
inaccuracy at high P levels even with the 30.5-m grid. Smith er al. (1998) suggest
that more intense grid sampling is needed to assess soil fertility in the cotton
(Gossypiurn hirsuturn) fields they studied. Once an accurate map of P and K lev-
els is created, then expectations about the relationship between yield and soil fer-
tility need to be kept in perspective. A low or lack of correlation between soil fer-
tility and yield in a given year means that other factors regulated yield. This
outcome does not diminish the value of following a proven soil fertility manage-
ment philosophy.
The low temporal variability in soil tests suggests that precision P and K man-
agement will work where fields vary spatially in their need for added inputs. Since
precision P and K management will ultimately lead to variable rate nutrient man-
agement based on replacing nutrients removed by crops, a precision management
program based on a proven soil fertility management philosophy will be success-
ful where crop yields vary in response to other unmanageable yield-limiting fac-
tors.
ASPECTS OF PRECISION AGRICULTURE 31

iii. NITROGEN.The precision management of N will be applicable to situa-


tions in which the factors that control total N in soils and N availability to plants
vary spatially (Pan er al., 1997). However, precision N management will be in-
creasingly more difficult but may in fact have increasingly more environmental
benefits as the temporal component of spatial variability of N availability in-
creases. For this reason, precision N management is more complex than precision
management of lime, P,and K but may have significantly more value. The diffi-
culty presented by a large temporal component of N availability is clearly illus-
trated by the data of Cahn et al. (1994). From May 13 to June 15, 1992, in a 1.1-
ha field, soil concentrations of nitrates in the 0 to 15-cm depth decreased on the
average from 8.3 to 1.6 mg kg-I and the spatial dependence observed in May was
no longer present in June (Fig. 4). Conversely, early success in precision N man-
agement has occurred where the temporal variability of a significant aspect of N
availability was low, a case exemplified by the success of precision N management
in sugar beets in the Red River Valley (Cattanach er al., 1996; Lilleboe, 1996) and
in irrigated corn in Nebraska (Hergert et al., 1996), where N leaching overwinter
was low and residual nitrates were high.
Total N in soils is primarily in the organic form (about 90%;Stevenson, 1982),
and most of the inorganic N is mineral-fixed N ammonium (Young and Aldag,
1982). Since total N in soil varies with soil organic matter, clay mineralogy, and
clay content of soils, and since N mineralization is the major natural source of in-
organic N in soil, then N in soil will vary spatially with soil texture and organic
matter content. Inorganic N, as nitrate and ammonium, is required by plants in
large quantities and its content in soil is intricately woven into the complex soil N
cycle processes of mineralization-immobilization, leaching, and denitrification
(Meisinger et al., 1992).In crop production, net mineralization rates from soil or-
ganic matter are generally 1 or 2%of total N (Meisinger and Randall, 1991), which
is often insufficient to supply the N requirements of crops that do not fix atmos-
pheric N. Therefore, N in the form of inorganic fertilizers or waste materials con-
taining N are applied to soil or a crop to meet crop N needs. More than 11.7 mil-
lion metric tons of commercial N fertilizers and more than l million metric tons
of N in animal manures were applied in the United States in 1995 (USDA-ERS,
1997).
Both N deficiency and excess N availabilitycreate problems for production agri-
culture. Most fertility concerns are focused on deficiencies in N availability to
plants because they reduce yield and/or quality of crops (Olson and Kurtz, 1982).
Excess nitrate-N in soil can lead to N losses to the environment that reduce water
quality (Mueller et al., 1995) and can reduce yield and/or quality for some crops.
Excess N increases protein content but reduces sugar content in sugar beet (Hills
and Ulrich, 1971) and oil content in canola (Brassica napus L.; Henry and Mac-
Donald, 1978), whereas both yield and crop quality can be reduced in wheat (Ras-
mussen and Rhode, 1991). Leaching, runoff, and denitrificationare processes that
32 FRANCIS J. PIERCE AND PETER NOWAK

May 1992

cn I
0, 15
a $ , h
a 2 o ~ m o i m
Y

v
P lo
5
z
' 0
0"
2

110

0 mw data

n
c
I
June 1992
EI, 1s
Y

U
F lo
5 a 20 40 60 w1100
z
1 0 Lag diatanco (m)
0
"
z

Figure 4 Spatial distributions and semivariograms of raw and residual NO,-N data (0-15 cm)
from the 1.1-ha area in May and June 1992 (reprinted with permission from Cahn era[., 1994).

result in loss of N from the soil-plant system creating the potential for N deficiency
in crops and degradation of water and air quality. The environmental threat of N
losses from soil is further compounded by the application of manures at rates in
excess of crop nutrient needs (Nowak et al., 1998). This problem is exacerbated
by the concentration of livestock and the high costs of transporting wastes result-
ing in excess animal manure being applied to limited areas (Hatfield and Stewart,
1997). Deficiencies and excesses of N can occur within the same field sometime
during the year. Thus, it is the variability in space and time of the processes that
ASPECTS OF PRECISION AGRICULTURE 33

regulate the availability of N to plants and the fate of N in soil that make precision
N management attractive.
Nitrogen management for crop production is to varying degrees prescriptive
by nature because crop yield is to a large extent determined by growing condi-
tions that occur after the N uptake phase in plants. Nitrogen management strate-
gies are approached in two basic ways: (i) prevention strategies, whereby pre-
scriptive applications of N inputs are made prior to or early in the N uptake phase
of plant growth to avoid nutrient deficiencies, and (ii) intervention strategies,
whereby N inputs are applied to meet N requirements as determined by the nu-
trient status of soils or plants during the rapid N uptake phase of growing plants.
Hybrid strategies supply a portion of crop needs early and intervene later. For
example, the early presidedress nitrate tests used in the northeast United States
were designed to adjust N fertilizer recommendations based on critical nitrate
concentrations in soil following preplant applications of fertilizer N (Magdoff et
al., 1990). Prevention is the most common N management strategy and is usu-
ally based on a mass balance approach in which fertilizer recommendations are
based on some combination of yield goal, N requirements of the crop, residual
soil N in the soil profile, and N mineralized from soil or plant residues (with
some using soil organic matter as a proxy for N mineralization). Intervention
strategies include foliar fertilization (Stone et al., 1996), delayed N applications
based on N content in plants (Schepers et al., 1992), and chemigation (King et
al., 1996) strategies that assess crop needs based on tissue sampling, sensing
plant reflectance, or crop simulation. From the current literature, it appears that
precision N management will be feasible using prevention strategies based on N
balance approaches that rely on soil testing for residual soil profile nitrates where
temporal variation in soil profile nitrates is low. Intervention strategies will be
more applicable to precision N management where the temporal component of
spatial variability is moderate to high. Intervention strategies that sense the oc-
currence of N deficiencies in plants during the uptake phase (Schepers et al.,
1996) may be better adapted to precision N management than strategies that use
mass balance approaches that are to a large extent soil sampling based, such as
the presidressed soil nitrogen test (PSNT), even where they may be useful on a
field level basis (Bock et al., 1992). The success of precision N management will
depend on how well we can predict and control the dynamic soil and crop
processes that regulate N availability in soils and plant N requirements (Pan et
al., 1997).
Research on precision N management to date has taken three basic approaches.
The first is based on the application of current N recommendations to site-specif-
ic within-field scales where some form of grid soil sampling is employed. Resid-
ual soil nitrate-N values are interpolated to generate N availability maps forming
the basis for N fertilizer recommendation maps (Cattanach et al., 1996; Ferguson
34 FRANCIS J. PIERCE AND PETER NOWAK

et al., 1996; Fiez et al., 1994a; Franzen et al., 1996; Hergert et al., 1996; Kitchen
et al., 1995; Redulla e l al., 1996). This approach is based on the prediction of N
requirement of crops using a balance sheet approach that is the basis for current N
recommendations for crop production (Meisinger et al., 1992) such as that de-
scribed by Eq. (1) as presented by Pan et al. (1997):
Nf = [ G , X 1I(NfINaVX NavlNs X Gw/Nt)]- (Nmin+ Nin) (1)
where Nf is the fertilizer N requirement, G, is the yield goal, Nav is the available
soil N in the root zone, N, is the plant N, N, is the predicted requirement for N sup-
ply to crop, Nminis the soil N supplied from net mineralization, and Ninis the pre-
plant inorganic N.
In practice, university recommendations for fertilizer N are usually much sim-
pler than that given in Eq. (l), basically a function of the product of yield goal
times an N factor adjusted for N credits and in some cases residual nitrates. The
tristate N fertilizer recommendation for corn in Indiana, Michigan, and Ohio is
based on yield goal and N credits from manure or legumes in rotation (Vitosh et
al., 1996), whereas Nebraska’s N recommendation for corn relies on three vari-
ables-yield goal, residual soil nitrate, and percentage organic matter as a proxy
for N mineralization (Hergert et al., 1995a). The second approach is to develop
site-specific optimal N rate recommendations based on condition-specific N re-
sponse curves (Blackmer and White, 1996; Kachanoski etal., 1996; Malzer et al.,
1996) andlor related to landscape attributes (Hollands, 1996; Solohub et al., 1996;
Vetsch et al., 1995).A common experimental procedure is to apply a range of N
rates in replicated strips across a field, including a no N treatment, to obtain an op-
timal N rate by condition within a field. Ultimately, the derived site-specific op-
timum N rates must be related to soil or landscape properties if they are to be
useful elsewhere. The third approach is to develop site-specific intervention N
management based on crop monitoring of N status (Bausch et al., 1996; King et
al., 1996; Schepers et al., 1992, 1996; Stone et al., 1996; Vetsch et al., 1996).The
idea is to monitor plant N concentration by monitoring plant or canopy reflectance
of light or some measure of plant N content such as chlorophyll content (Schep-
ers et al., 1992), estimate N fertilizer requirement using established relationships
between reflectance and N content, and fertilize the crop to the optimal N content
for maximum economic yield. In some cases, a portion of a crop is fertilized to op-
timal levels and used as the standard for adjusting the N recommendationsfor the
remainder of a field (Schepers et al., 1996) or plant N is estimated using a previ-
ously developed index calculated from measured canopy reflectance (Bausch et
al., 1996). Some success has been attained with each of these approaches but none
are broadly applicable.
Where and when will precision N management pay off in terms of either prof-
itability or environmental benefits? Since the key to precision N management is
that the factors that regulate these processes vary in space and time we expect pre-
ASPECTS OF PRECISION AGRICULTURE 35

cision N management to be profitable and/or beneficial to the environment to vary-


ing degrees under the following conditions:

Where N inputs are high: Dryland wheat would be marginal due to low inputs
of N and low yields (Cam et al., 1991 ; Solohub et al., 1996), whereas high-yield
wheat production in the Pacific Northwest would have potential because of high
N inputs, high residual N, and high yields (Fiez et al., 1994a,b, 1995). Corn ac-
counts for the major input of fertilizer N and use of manure, whereas potatoes
(Solanurn tuberosurn L.) receive the highest average N application rate of all
crops (248 kg N ha-'; USDA-ERS, 1997). These crops are the focus of much
of the current research on precision N management.
Where residual N is temporally stable and/or high residual N is predictable
from the yield of the previous crop, for example, low yields the previous year:
Residual N is reported to be spatially variable and stable from fall to spring in
irrigated corn production in Nebraska (Hergert et al., 1995b) and in soil pro-
files prior to sugar beet production in the Red River Valley of eastern North
Dakota and western Minnesota (Cattanach et al., 1996). Residual profile N is
frequently leached in the humid regions over winter (Everett and Pierce, 1996;
Hoeft et al., 1992), except where climatic factors, such as low annual precipi-
tation and temperatures that keep soils frozen for extended periods overwinter,
favor retention of residual profile nitrates (Bundy et al., 1992). Residual profile
nitrates are not considered in nitrogen recommendations for corn or wheat in
certain areas such as the tristate area of Indiana, Michigan, and Ohio (Vitosh et
al., 1996).The PSNT, which is primarily an index of mineralization in the sur-
face 30 cm of soil, may be relevant in some humid regions but not others (Bock
et al., 1992).
Where crop quality is affected by excess N is soil: A major benefit of grid soil
testing and variable rate N application is a significant increase in sucrose con-
tent in sugar beet (Cattanach et al., 1996; Lenz, 1996; Lilleboe, 1996).
Where crop yield spatial variability is high and predictable: Nitrogen recom-
mendations are based primarily on crop yield. If yield is variable, then precision
N management based on yield goal makes sense if the yield variation is pre-
dictable from year to year (low temporal variability). If yield variability is high
but not predictable, then precision N management will be difficult because it is
generally prescriptive. At this time, it is common to use a uniform yield goal in
site-specific N fertilizer predictions while varying N recommendationsbased on
residual N (Long et al., 1996) and in some cases organic carbon (Hergert et al.,
1996).There have been some attempts at using previous-year yield maps to pre-
dict N requirements, although previous yield has not been very useful (Kitchen
et al., 1995) and some suggest previous yield maps have no relevance (Vetsch et
al., 1995).Some suggest that previous-year yield maps do not reflect N require-
ments because other factors besides N availability regulate yields (Kachanoski
36 FRANCIS J. PIERCE AND PETER NOWAK

et al., 1996; Malzer et al., 1996). Yields have been related to spatial variability
of water availability associated with landscape position (Mulla et al., 1992).The
use of digital terrain models to predict water availability in the landscape is
promising (Bell et al., 1995).
Where net mineralization is high and consistently related to soil and landscape
properties: Soils with low residual nitrates from the previous crop can provide
a significant portion of the N requirements of the succeeding crop through min-
eralization of soil organic N and crop residues. Pan et al. (1997) calculated that
mineralization of N from the surface 30 cm ranges from 39 to 224 kg N ha-'.
Vetsch et al. (1995) report that the inability of current fertilizer recommenda-
tions to predict soil N supply and fertilizer N use efficiency in the field makes
them inadequate to capture the benefits of site-specific N management. Walters
et al. (1996) predicted N requirements from soil organic matter maps. Steven-
son and van Kessel (1996) showed that following pea in rotation, the depres-
sional areas had higher soil N and a reduction in fertilizer N would be needed
in wheat to reduce the negative effect of excessive soil N supply, but this was
not the case following wheat in rotation. Corn following legumes in rotation
can have little response to fertilizer N (Frye et al., 1988; Hesterman, 1988).
However, spatial variation might be expected in legume contributions within a
field.
Where N application is not restricted in time: Sprinkler-irrigatedlands have un-
restricted accessibility to the field to apply N based on crop needs using chemi-
gation (Evans et al., 1996; King et al., 1996). Stone et al. (1996) used spectral
radiance measurements for correcting in-season wheat N deficiencies at Feekes
physiological stages 4 and 6.
Where leaching potential is high and spatially variable prior to or during the crop
N uptake period of plant growth: Within-season leaching events result in losses
of nitrate either through leaching itself or due to denitrification if soils remain
saturated over extended periods. Because loss of N may vary with landscape po-
sition and soil variability, intervention applications of N inputs to reduce N de-
ficiencies could be varied spatially. On the preventative side, anticipation of ar-
eas where such losses might be expected is the basis for variable rate application
of nitrification inhibitors (Malzer et al., 1995).
Where variation in topographic position regulates N availability or yield: Cer-
tainly yield is very much dependent on water availability-whether drought or
in excessive amounts. Effects of landscape position on wheat yields have been
reported by Fiez et al. (1994a,b, 1995) and Halvorson and Doll (1991). In the
Red River Valley, N availability is higher in knolls than in depressions and is key
to managing variable rate N in sugar beet (Holland, 1996; Franzen et al., 1996).
Techniques to predict water in the landscape have been enhanced by recent ad-
vances in digital terrain analysis (Bell et d., 1995; Moore et d., 1993). Thus,
ASPECTS OF PRECISION AGRICULTURE 37

where soils or landscapes consistently regulate water availability, precision N


management would have potential.
On the negative side, precision N management will not work well under the fol-
lowing conditions:
Where the previous crop exceeds yield goal: When yield exceeds yield goal as
it did in 1994 (Hergert et al., 1995b), residual nitrates were reduced. Since the
benefits of grid sampling appear to increase with increasing residual N levels and
increasing spatial variability (Lenz, 1996),higher than expected crop yields may
diminish the prospects for variable rate N success the following year.
Where large-scale leaching events occur prior to the growing season: In the ar-
eas flooded in 1993, additional fertilizers, including N, were applied to replen-
ish flood-damaged soils (USDA-ERS, 1997). Since nitrates are soluble, large
leaching events should minimize residual nitrates within the root zone.
Where precision N management has successfully reduced spatial variation in
yield and residual N: The short-term financial benefit of precision N manage-
ment is an increase in N use efficiency by accounting for residual N in the N fer-
tilizer recommendation, thereby reducing N fertilizer as well as reducing resid-
ual N available for leaching (Hergert et al., 1995b).Therefore, the benefits from
precision N management are a result of accounting for residual N that is spatially
variable. As farmers become skilled in this technique, the long-term effect of pre-
cision N management is a reduction in residual N, making precision N manage-
ment less profitable although N leaching is continually reduced. Farmers do not
get paid for environmental benefits so good precision management will lose its
potential profit benefit to farmers. It appears that when residual N is low, grid
sampling and VRT will be less profitable. On the other hand, if precision N man-
agement is to increase soil N to maximize economic yield where inadequate
amounts of N would be applied under uniform management, then residual N will
be higher than it would with uniform N management, unless N use efficiencies
increase proportionately. Actually, residual N will be an issue when yields from
the previous crop are less than yield goals and/or nitrogen use efficiencies are
reduced. Therefore, yield maps combined with grain protein content and with N
application maps should be a guide to precision N management but only where
overwinter leaching is low.
Where there is a strong temporal component of spatial variability: Where resid-
ual N is low due to high leaching potential, precision N management must rely
on prediction of crop yield, net N mineralization, N losses during N uptake pe-
riod, and where amendments such as nitrification inhibitors will synchronize ni-
trate availability with crop uptake demand. Not only does the prediction of the
spatial variability of inorganic N over time make precision N management much
more difficult but also we believe the potential to the farmer and the environ-
38 FRANCIS J. PIERCE AND PETER NOWAK

ment are larger than when temporal variation is higher. The ability to use inter-
vention N management strategies may allow fanners to overcome the manage-
ment difficulties when the availability of N encountered has a large temporal
dependence.
The potential for precision N management can also be considered in terms of
the steady-state and non-steady-state approximations to whole-crop N balance in
the soil-crop system (Table I) discussed by Meisinger and Randall (1991). Soil-
crop systems that are steady state will have fertilizer N needs that are directly pro-
portional to crop removals and indirectly proportional to fertilizer efficiency be-
cause no soil N availability terms appear in the steady-state case. Situations con-
ducive to steady state, such as those listed in Table I, are common in humid regions
(Meisinger et al., 1992). Precision N management for the steady-state case would
be based on crop yield variability and fertilizer use efficiency, both related to soil
and landscape properties. Fields with dissimilar soils and/or variable landscapes
would be well suited to precision N management if crop yield variability were
known or predictable, whereas fields with similar soils and uniform landscape
would not. Soil-crop systems that are non-steady state have high temporal vari-
ability associated with management, soil, and climatic factors. Such conditions are
encountered with crop rotations involving forages, organic inputs, changes in
drainage, or tillage or with large climatic fluctuations. Precision N management
may be easiest when climatic factors favor N accumulation in soil profiles (low
precipitation) but more difficult when management or climatic factors result in N
transformation processes that are dynamic. An evaluation of the steady- versus
non-steady-state approach to precision N management needs to be performed.
In summary, the potential for precision N management is directly related to the
extent of spatial variability in the factors that regulate N availability in soils,
whereas the difficulty in precision N management is related to the degree of tem-
poral variability. Where spatial variability is high and temporal variability is low,
precision N management appears to be profitable and to reduce N available for
leaching. Intervention N management strategies will be needed where the tempo-
ral component of N availability is high but access to the crop will be required for
delayed N fertilizer applications. Finally, residual nitrates are human induced. As
precision N management begins to control the anthropogenic sources of variation,
only the natural sources will remain to be dealt with.

b. Precision Pest Management


Weeds, insects, and diseases are an ever-present and costly management prob-
lem to crop production because significant infestations reduce crop yield and/or
quality and, if severe, can limit crop production options. The USDA-ERS (1997)
reports annual expenditures of $7.5 billion on agricultural pesticides in the Unit-
ed States, of which two-thirds is spent on herbicides and about one-third on in-
ASPECTS OF PRECISION AGRICULTURE 39

'hble I
Characteristics of Steady-Statevs Non-Steady-StateMi-Crop Systemtf
~~ ~~ ~~

Circumstances conducive to Circumstances conducive to


steady-state approximation non-steady-state approximation

Goal or objective
Objectives of N budget lie in making estimates Objective of N budget is for a short-term period,
over a long-term period (10 year average) e.g., the next crop (1-year period)
Management factors
The same soil and water management system in Arecent (within 1-5 years) change in the soil
place for an extended period (5-20 years), and water management system, e.g., new
e.g., same tillage system, same soil drainage tillage system and new drainage or irrigation
system, and same irrigation system system
The same crop management practices used for A recent change in crop management, e.g., new
an extended period, e.g., same rotation and rotation system and recent use of cover crops
long-term use of cover crops
The same N management system practiced for Arecent change in N management practices, e.g.,
an extended period, e.g., same N source recent use of manure to replace fertilizer N
(manure or fertilizers)and same timing and
placement practices
Soil and climatic factors
Soil total N content approximately constant over Soil total N content changing systematically
a long period (10-20 years), e.g., implies that over time, e.g., decline in soil N when grass-
N mineralization approximates N in residues land is plowed and increase in soil N with
plus immobilization continued manure applications
Soil inorganic N content approximatelyconstant Soil inorganic N content changes widely from
over time step, e.g., soil leached out each year, year to year, e.g.. periodic drought and
highly permeable soil in humid region, and irregular irrigation
excess irrigation
Climate (precipitation and temperature) not Climate highly variable and slows organic N
highly variable and is conducive to organic N turnover, i.e., cool and marked cold season or
turnover, i.e., warm and humid and predictable dry season and unpredictable hydrologic
yearly hydrologic cycle cycle
Soil type has low total N content, low NO, Soil type has high total N content, readily retains
content, shallow root zone, well-drained, high NO,, deep root zone, poorly drained, low infil-
infiltration,deep water table, low clay content, tration, shallow water table, high clay content,
e g , sandy loam soil in warm humid climate e.g., a clay loam soil in a cool semiarid climate

"Reproduced with permission from Meisinger and Randall (1991).

secticides.More important, public concerns regarding the impacts of pesticide use


include health risks related to food safety, water quality, and worker safety and
concerns over wildlife and ecosystem health. Therefore, agricultural management
practices that reduce pesticide use, improve pest management, or reduce risks of
pesticides to human and ecosystem health are very desirable.
The intuitive appeal of precision agriculture is that it offers the potential for such
40 FRANCIS J. PIERCE AND PETER NOWAK

benefits. The potential direct economic benefit of precision pest management to


farmers is a reduction in chemical/nonchemicalpest management costs, crop dam-
age, or both due to more efficacious or efficient application of pest control mea-
sures. A reduction in pesticide use, however, does not translate into profit if the
cost of obtaining information about pest populations and distribution exceeds the
savings (Forcella, 1993; NRC, 1997).Environmental benefits are presumed to re-
sult from a reduction in pesticide usage, particularly in sensitive environments,
although the results of the few studies on the potential of precision farming to
provide environmental benefits have been inconclusiveregarding its effect on pes-
ticide use (USDA-ERS, 1997). There are two basic questions in precision pest
management: Is it possible to effectively manage pests spatially? and Will it pay?
Here, we assume that spatial precision pest management is doable so that we
can consider the potential benefits of it. If the main benefit of precision pest man-
agement is a reduction in input costs and applications and avoidance of unneeded
applications for environmentalbenefit (NRC, 1997),then the task at hand is to de-
termine where pesticides are used and then assess the potential for precision man-
agement. A summary of pesticide use in the United States by selected crop is giv-
en in Table 11. Four crops account for 78% of pesticide use on major crops in the
United States [corn (35.6%), potatoes (15.4%), cotton (14.8%), and soybean
(12.1%)]. Vegetable crops account for approximately 12%. Corn, soybean, and
cotton account for 89% of the herbicide use in the United States. Cotton, corn, and
potatoes account for 69% of the insecticide use, and fruits and vegetables account
for 29%. Potatoes and vegetables account for 66% of the fungicide use, with fruits
accounting for 30%. Potatoes and cotton account for 73% of other pesticide usage,
a category which includes soil fumigants, growth regulators, desiccants, and har-
vest aides; vegetables account for 26%. If savings in chemical uses is sought, then
we need to examine where the chemical use is taking place.
Usage does not necessarily imply potential benefit from precision management.
The average annual pesticide costs for corn, soybean, and cotton are 62, 59, and
124 $ ha-' compared to 15 $ ha-' for wheat (USDA-ERS, 1997). Consider a
spraying operation costs about $5 ha-'. If a farmer wants input cost and applica-
tions savings to generate profits from precision management, then the fraction of
input and application costs saved by precision management must exceed the cost
of information gathering and variable input application costs. Crop improvement
by improved pest management would add to the profit, and crop damage and/or in-
tervention costs for pest escapes from inadequate management would add to the
cost. The risks and associated codbenefit for either scenario have not been as-
sessed. For crops that have low annual pesticide costs, precision pest management
will be hard to justify economically solely on pesticide savings given the potential
cost of precision management. Crop improvement or returns for environmental
benefits from precision management will be needed. Currently, farmers are not
compensated for environmental benefits accrued from improved pest management.
Table II
Pesticide Use in 1995 in Selected U.S. Crops by Pesticide

kg ai %

crop Herbicide Insecticide Fungicide Other Total Total Herbicide Insecticide Fungicide Other

Corn 84,587 6,790 9 0 91,385 35.6 57.5 21.5 0.0 0.0


Cotton 14,924 13,638 474 8,959 37,995 14.8 10.2 43.2 2.3 15.5
wheat 9,105 413 227 0 9,745 3.8 6.2 1.3 1.1 0.0
Sorghum na na na na 0 0.0 na na na na
Rice na na na na 0 0.0 na na na na
soybeans 30,929 234 6 0 31,169 12.1 21.0 0.7 0.0 0.0
Peanuts na na na na 0 0.0 na na na na
Potatoes 1,314 1,411 3,620 33,109 39,454 15.4 0.9 4.5 17.8 57.2
Other Vegetables 2,778 2,530 9,902 15,115 30,325 11.8 1.9 8.0 48.7 26.1
Citrus 2,118 2,335 1,825 81 6,359 2.5 1.4 7.4 9.0 0.1
Apples 348 1,618 2,125 42 4,133 1.6 0.2 5.1 10.4 0.1
Other Fruit 898 2,628 2,154 554 6,235 2.4 0.6 8.3 10.6 1.o
Total 147,001 3 1,598 20,341 57,860 256,800

"Adapted from ESRI (1997).


includes soil fumigants, growth regulators, desiccants, and harvest aides.
42 FRANCISJ. PIERCE AND PETER NOWAK

Therefore, it appears that the potential for benefits from precision pest manage-
ment, assuming it is doable, depends on the dependence on a particular crop for
costly pest management practices (e.g., pesticide use) and actual improvementsto
crop yield, crop quality, and/or to the environment from improved pest manage-
ment practices. There are risks associated with being wrong, and these should en-
ter into the discussion because farmers understand that they are managing risks in
producing a crop. Next, we examine the potential for precision management for
weeds and insects. There is little work on precision management of diseases re-
ported in the literature. Diseases vary spatially within a field but are difficult to
predict since they are host specific and vary temporally.

i. WEEDMANAGEMENT. The application of precision agricultureto weed man-


agement is potentially beneficial to agriculture because (i) it offers an opportuni-
ty to reduce chemicalhonchemical inputs into crop production through site-
specific weed control and the use of precise application techniques and (ii) the
acquisition of spatial and temporal information on weed occurrence and distribu-
tion made possible with precision agriculture technologies will lead to an im-
proved understanding of weed biology and ecology needed to develop more ef-
fective weed management strategies (Johnson et al., 1997). Precision weed
management is possible because weeds are spatially aggregated and not random-
ly distributed within most agronomic fields (Dessaint et al., 1991; Johnson et al.,
1995; Mortensen et al., 1993, 1995, 1998; Von Groenendael, 1988) and because
the efficacy, efficiency, and fate of weed control inputs vary with weed and crop
conditions and with soil physical and chemical properties, all of which can vary
spatially and, to a varying extent, temporally (Johnson et al., 1997). It is impor-
tant to understand that precision weed management is much more than variable
rate application, although this aspect receives the bulk of attention. Precision weed
management also involves an understanding of the spatial and temporal interac-
tions between landscape characteristics, pest populations, and weed management
strategies; this understanding is currently lacking (Johnson et al., 1997). Consid-
er that technologically it will be possible to perform field operations very precise-
ly at night when environmental conditions may be more conducive to weed man-
agement operations, such as spraying in the dark to take advantage of cooler
temperatures and reduced winds (Holmberg, 1998). Research has already shown
that night tillage can be effective in control of the germination of weed species for
which germination is light induced (Asgard, 1994; Hartmann and Nezadal, 1990;
Scope1 et al., 1994). Thus, a very important dimension of precision weed man-
agement is that new weed management strategies will evolve because of it.
While not the only aspect of precision weed management, its potential for re-
duction in herbicide use may be most important in its adoption because of public
concerns regarding the environment and food safety. Management systems and in-
puts that foster a reduction in herbicide use in agriculture are very desirable for
ASPECTS OF PRECISION AGRICULTURE 43

economic and environmental reasons, and these aspects have been an important
focus in crop protection research. Pesticide product formulations have changed to
lessen environmental and human health effects, to reduce the development of pes-
ticide-resistant pests, and to provide more cost-effective pest controls (USDA-
ERS, 1997). Application rates have been lowered for new herbicides, which re-
quire a fraction of the previous rates. There has been a shift from preemergence to
postemergence herbicides, using chemicals that have greatly reduced soil residual
activities, that foster integrated pest management (IPM), that require lower rates
when applied to small weeds, and that when combined with herbicide-resistant
crops lower herbicide costs and amounts. These newer herbicides and crop pro-
tection technologies will benefit from precision application techniques, further
supporting the case for precision weed management. However, the results of the
few studies on the potential of precision farming to provide environmental bene-
fits have been inconclusive regarding its effect on pesticide use (NRC, 1997;
USDA-ERS, 1997).
A common approach to precision weed management is site-specific weed con-
trol achieved by (i) applying herbicides only where weeds are present or above
economic threshold levels, termed intermittent herbicide application (Mortensen
et al., 1995) or patch spraying (Stafford and Miller, 1993, 1996); (ii) varying her-
bicide application (type, formulation, or rate) according to soil physical and chem-
ical properties or weed characteristics (species, growth stage, and density); or (iii)
some combination of the two approaches (Johnson et al., 1997). For prevention or
preemergence weed control, site-specific application requires prior knowledge of
historical weed distributions since no weeds are visible at the time of application.
This knowledge can be obtained by mapping weed aggregation in previous years
(Brown et al., 1990; Lass and Callihan, 1993).Weed control treatment or intensi-
ty can also be varied based on soil properties according to label or other recom-
mendations if knowledge of the spatial variation of these properties is adequately
known. Clearly, however, the county soil survey is generally not sufficient in de-
scribing the variability in surface soil properties affecting herbicide performance
(Mausbach et al., 1993). Therefore, this precision weed management strategy re-
quires intensive and expensive soil variability assessment.
Intervention or postemergence weed control treats emerged weeds and inter-
mittent or patch-spraying weed control systems require either a map of weeds pres-
ent just prior to a weed control application or a real-time sensing of existing weeds
at the time of application.The key is that weeds are aggregated into patches rather
than randomly distributed. Weed mapping has been performed using remote sens-
ing (Hanson et al., 1995). real-time ground-based detection using light reflectance
(Duff, 1993; Felton et al., 1991; Felton and McCloy, 1992; Haggar et al., 1983;
Nelson, 1993; Shearer and Jones, 1991; Shropshire et al., 1990) or digital image
processing (Guyer et al., 1986; Sadjadi, 1996; Woebbecke et al., 1995), and GPS-
assisted scouting (Colliver et al., 1996; Stafford et al., 1996; Stafford and Miller,
44 FRANCIS J. PIERCE AND PETER NOWAK

1996).There are limitationsto real-time systems in weed detection, in sprayer con-


trol, and in controlling herbicide type or concentration in real time. These include
difficulty in separation of weeds from the crop, the time lag between weed detec-
tion and weed control application, spray pattern effects caused by the transport de-
lay between change in herbicide type or rate and chemical discharge at a given
nozzle, and nozzle performance as affected by herbicide type and rate change al-
terations of droplet size and flow rate for a given nozzle design or nozzle wear (An-
derson and Humburg, 1997). Additionally, equipment size (boom width and noz-
zle spacing) will determine the scale of weed management zones, with larger
equipment spraying larger areas relative to the nonweed areas (Johnson et al.,
1997).Extensive research on sprayer control, weed detection, and spot spraying is
under way (Ballal et al., 1996; Giles et al., 1996; Kirk and Tom, 1996; Nuspl et
al., 1996).
Regardless of the approach, the economic and environmental benefits of site-
specific herbicide application are derived from applying herbicides only to areas
occupied by weeds at rates adequate for weed control and using environmentally
safe herbicides or rates. The smaller the area occupied by weeds, the larger the
potential benefits. The key, therefore, is to know the spatial distribution of weed
populations. Lack of knowledge of weed distributions results in over- or underap-
plication of herbicides or other weed control measures-errors which have asso-
ciated costs. The extent of overapplication(application to areas that did not require
it or overapplication for the conditions present) is difficult to measure unless over-
application caused measurable injury to the crop. Underapplication, on the other
hand, results in weed escapes that are quite visible and often lead to yield reduc-
tions and weed-induced harvest problems. The risk of underapplication of weed
control practices, because it is most visible, will drive farmers’decisions more than
potential environmental concerns of overapplication. The potential for weed es-
capes associated with inadequate knowledge of weed distributions or inability to
control inputs sufficiently probably limits the adoption of precision weed man-
agement techniques.
Johnson et al. (1997) provide some generalities with regard to precision weed
management and we summarize them here. Weeds are aggregated within fields but
are more aggregated at low populations than at high ones. However, the best eco-
nomic returns to simulated site-specific weed management were obtained when
weed pressure and aggregation were high. Weeds more costly to control and her-
bicides with high costs or with high environmental sensitivity favor the use of pre-
cision weed management. Therefore, inexpensive, low-hazard chemicals are less
economical for precision weed management than expensive, hazardous chemicals.
Use of economic threshold values increases the amount of area within a field that
does not require herbicide. Economic thresholds were developed for uniform her-
bicide management and do not include the effects of aggregated weed distribu-
tions. Site-specific weed management may be more important for early emerging
ASPECTS OF PRECISION AGRICULTURE 45

weeds than later emerging weeds since crops are more competitive later in the sea-
son. Stability in weed patches will affect the ability to predict weed distributions
that do occur in localized patches, with weeds in no-tillage more stable than those
in tilled systems. In the future, site-specific weed management may shift weed
species by selecting for weeds with long seed dispersal mechanisms (nonpatchy
distributions),with light reflectance properties or morphologies similar to those of
crops that escape preemergence herbicides or that build resistance to herbicides in
programs that use pre- and postemergence herbicides with the same mode of ac-
tion. On the other hand, improved knowledge of the factors driving weed presence
or absence, whether they be management, biological, or environmental, should
lead to better understanding of weed biology and ecology which should lead to
better weed management strategies. This aspect of precision weed management is
yet to be fully explored.
Ultimately, the extent of herbicide reduction in precision weed management is
contingent on the weed infestation level, spatial distribution, soil heterogeneity,
and performance of the application equipment. Forcella (1993) suggests that man-
agement of spatial variability is worthwhile as long as the degree of variability is
large enough to justify the cost of obtaining the information and managing the dif-
ferences accordingly. The growing body of evidence from research studies on in-
termittent or patch weed control suggests that considerable areas within fields are
weed free or have weed densities below economic threshold levels (Colliver et al.,
1996; Gerhards et al., 1996; Heisel et al., 1996; Johnson et al., 1995; Mortensen
et al., 1995; Sadjadi, 1996; Stafford and Miller, 1996), but a thorough assessment
of the economic and environmental benefits from these practices has not yet been
done. The risk of weed escapes may also play a role in adoption patterns but this
has not been addressed in most studies. Stafford and Miller (1996) conclude that
methods for automatic detection of weeds have yet to be developed to the point
where they can be used in production agriculture. Therefore, research should fo-
cus on the acquisition of knowledge and understanding of the spatial distribution
of weeds and the benefits to sustainable weed management that can be derived
from an increased understanding of the distribution.

ii. INSECTMANAGEMENT. Precision insect management has potential be-


cause distributions of insect populations are spatially variable, in part because in-
sects are mobile during at least part of their life cycle and in part because during
the relatively nonmobile stages insects cluster in response to environmental (e.g.,
temperature and moisture) and behavioral responses (Heischer ef al., 1997; Tay-
lor, 1984).Therefore, precision insect management has the potential to reduce in-
secticide applications and improve the efficacy of both prevention and interven-
tion insect management strategies. A major difficulty with the management of
insects is that their populations are highly dynamic and prediction of insect densi-
ty is difficult or uncertain, both of which make it necessary to collect field esti-
46 FRANCIS J. PIERCE AND PETER NOWAK

mates of pest density to monitor density over time (Fleischer et al., 1997).The dif-
ficulty and costs associated with repeated sampling of the spatial distribution of
insect densities are considered by some to be an insurmountable barrier to the use
of precision insect management in IPM programs (Fleischer et al., 1997). This dif-
ficulty is reflected in the limited research on the spatial management of insects. At
the precision agriculture conference held in Minnesota in 1996 (Robert et al.,
1996). only 1 of the 147 papers included in the proceedings dealt with insects, a
spatial characterizationof corn rootworm (Diabrotica sp.) populations in corn by
Ellsbury et al. (1996).
The review of Fleischer et al. (1997) represents the major synthesis on this top-
ic and is recommended reading. Beyond recognizing the sampling difficulties, they
propose many important benefits of precision insect management worth repeating
here. First, precision insect management is an enhancement of IPM in that it is a
continuation along the path of using knowledge to replacing chemical inputs. A
map of insect density is a more realistic model than a mean estimate of pest den-
sity currently used in IPM and may stimulate the use of IPM by farmers. Second,
the alarming increase in resistance requires management strategies that maintain
susceptiblephenotypes and ensure gene flow among susceptible and resistant sub-
populations.This need, coupled with the need to maintain predator and parasitoid
populations, supports the notion of leaving habitats untreated within fields, areas
referred to as refugia. The creation of temporally dynamic refugia within fields is
a new development in insect IPM and is a capability of precision agriculture.Third,
the improved knowledge provided by precision insect management regarding in-
sect populations across a landscapecan be used to enhance areawide and landscape
pest management programs.
Currently, the sampling problem inhibits progress in precision insect manage-
ment. Commercially acceptable sampling schemes are not available and it is un-
known which insects and which crops will respond to precision insect manage-
ment (Fleischer et d., 1997). Like others dealing with other aspects of precision
agriculture, Ellsbury et al. (1996) conclude their paper with the caveat about the
need to know variability in that precision insect management should lead to re-
duced pesticide inputs “provided the spatial variation of rootworm populations can
be economically and reliably monitored and predicted.” Despite sampling diffi-
culties, positive effects of site-specific management have been reported. The de-
velopment of new sampling schemes and technologies for rapid scouting of fields
should create opportunities to develop new insect management interventions that
are economical and environmentally friendly.

c. Crop Management
The potential for precision crop management derives from the genotype X en-
vironment interaction (G X E). The G X E refers to changes in the relative per-
formance of cultivars across different environments. where environment includes
ASPECTS OF PRECISION AGRICULTURE 47

all variables encountered in producing a crop, including soil type, soil fertility,
moisture, temperature, and cultural practices (Fehr, 1987). Precision crop man-
agement could be achieved by varying cultivars and/or by manipulating planting
geometry (populations, spacing, and seeding depth) according to variation in en-
vironments within fields. Varying cultivars or planting geomebies within fields has
potential if the G X E is such that the rank among genotypes changes across en-
vironments but not if the differenceamong genotypes varies without any alteration
in rank. The success or precision crop management, therefore, depends on whether
cultivars are available that have been bred for adaptation to specific environments
rather than adaptation over multiple environments. The technologies needed to
vary cultivar or planting geometry are already available (Anderson and Humburg,
1997), but the agronomic basis for varying them is not clear.
Farmers often plant more than one variety within a field and yield maps have
revealed that a G X E interaction can frequently exist within fields. The value of
varying cultivars within fields is illustrated by comparing the performance of two
cultivars along parallel swaths within a field (Fig. 5). At some locations along the
transect, the two cultivars do not vary, whereas at some locations cultivar A per-
forms better than cultivar B or B outperforms A. The obvious management strat-
egy is to plant each cultivar where it performs best and either cultivar where there
are no performance differences. While intuitively appealing, there are three major
limitations to variable cultivar management. First, of critical importance is the ex-
istence of cultivars that are responsive to specific environments. Many cultivars
are bred to perform best over a range of environments, although yield mapping has
revealed that this may not be as true for current cultivars as was previously be-
lieved. Criteria for variable cultivar selection may be based on risk management
rather than overall performance. For example, cultivars with higher resistance to
a specific stress may be preferred over less resistant cultivars even if under low
stress the latter cultivar performs considerably better. Second, cultivar selection is
based on expectations of past performance and there may be few data available to

Means A = B

Distance
Figure 5 Illustration comparing the performance of two varieties dong parallel swaths across the
length of a field.
48 FRANCIS J. PIERCE AND PETER NOWAJC

quantify the environmentunder which cultivar performance was evaluated. There-


fore, there may be a limited basis on which to base within-field cultivar selection
decisions. Finally, there may be limited knowledge about the spatial variation of
environments within fields or the temporal variability of within-field environments
is sufficient to make prediction difficult. Precision farming technologies do make
it possible to evaluate cultivar performance by environment through precision
planting and yield mapping. By collecting and analyzing yield maps documenting
cultivar performance by environment across a region, the agronomic basis for
varying cultivars within fields could be developed (Peterson, 1997). One problem
may be that the life span of a commercial cultivar may be too short to develop
site-specific recommendations. Conversely, the intensive on-farm evaluation of
cultivars may accelerate the collection of cultivar performance data allowing for
earlier release of improved cultivars or improved screening of poor performing
cultivars.
Varying planting geometry within fields has potential because plants respond to
competition for light, water, and nutrients, and competition is reduced when spa-
tial arrangement of plants is improved (Sojka et al., 1988). For plant distribution
to be a limiting factor to crop yield, other limiting factors need to be eliminated
(Porter er al., 1997). Sojka et al. conclude that a yield advantage may exist for a
wide range of species if established in dense, uniformly spaced canopies than for
more open canopies, provided early weed management, water availability,and fer-
tility are adequate. The caveat suggests that within-field variability of environ-
ments might support variable planting geometries.
The evidence specifically supporting variable seeding rates is limited and in-
consistent. Bullock et al. (1998) reported on an extensive variable seeding rate
study in corn at 170 individual locations across the corn belt from 1987 to 1996.
They found that economically optimal plant densities and field quality were cor-
related. However, they concluded that variable rate seeding will only be profitable
to the farmer who has sufficient knowledge about the relationship between yield
and plant density for each section of his fields, far more knowledge than any farmer
currently possesses. Others have shown positive responses to variable seeding
corn. Barnhisel et al. (1996) reported higher corn yields from variable seeding
rates based on topsoil depth, which greatly influenced yields at this location. They
cite yield increases from variable seeding corn in Indiana of 0.125-0.878 mg ha-'
as reported by Reichenberger (1996). Fiez and Miller (1995) evaluated seeding
rates for winter wheat by landscapeposition in the Palouse region of eastern Wash-
ington State. They reported yield averaged over two site years increased 10.3%
when seeding rates were doubled on the north backslope position but not on oth-
er landscape positions. Increased seeding costs were more than compensated by
increased yields. Since the yield increases were attributed to increased spike den-
sity, they suggest that cultivars with greater tillering potential and adaptation to
cooler, lower light conditions on north backslopes might result in similar yield in-
ASPECTS OF PRECISION AGRICULTURE 49

creases without increasing seeding rate. They refer to the results reported by Ciha
(1984) that indicate soft white winter wheat cultivars significantly vary by land-
scape position and there was a cultivar-landscape position interaction for spikelets
per plant and 1000-kernel weight.
While not well documented, a potential for variable seeding rates is suggested
by research on planting geometry. Interest in increasing plant populations in corn
is fostered by the fact that modern corn hybrids tolerate higher plant density stress
more than older hybrids (Tollenaar, 1991).This may be due in part to the fact that
newer hybrids appear to have improved ability to resist barrenness and other types
of injury associated with above optimum plant populations (Nafziger, 1994). Cox
(1996) found a hybrid-plant density interaction for dry matter and grain yield in
corn, suggesting that modern commercial corn hybrids interact with plant density,
regardless of growing conditions. They state, however, that some modem hybrids
do not tolerate high or even moderate plant density stress in dry years. A hybrid-
density interaction would mean that some varieties would respond to variable
seeding rates while others might not, depending on the stress conditions within a
field. Thomison and Jordan (1 995) evaluated the effect of hybrid differences in ear
growth habit and prolificacy and concluded they are of limited importance in de-
termining optimum plant populations compared to environment, hybrid, and plant
population main effects. Nafziger (1996) reported the net effect of doubles (two
plants growing in the same space) was to increase corn yield, whereas the net ef-
fect of skips (missing plants) was to decrease corn yield, both resulting primarily
through their effects on plant population. Plant spacing variability may increase
with planting speed (Nielsen, 1995) or with soil conditions that affect plant stands.
Therefore, where planting operations or soil conditions increase the stand loss
(more skips) through poor germination or emergence, increasing the plant popu-
lation in those locations in anticipation of reduced stands may increase yields over
uniform seeding rates. This is consistent with Nafziger’s (1996) suggestion that
the primary strategy to overcome the effects of skips in the row may be to increase
the anticipated plant population. Gaps within rows can be more important in de-
termining yield in soybeans than plant population differences when stand densi-
ties are moderate to high (Hicks et al., 1990). Devlin et al. (1995) evaluated the
influence of environment on the optimum row spacing and seeding rate for soy-
bean in Kansas from 1991 to 1993. At all 11 locations, soil fertility was adequate
and rainfall levels and planting dates covered the range of conditions expected in
Kansas soybean production. They concluded that narrow rows should be recom-
mended for locations where adequate moisture and high yields (>50 bushels per
acre) are expected and wide rows recommended where soil moisture is expected
to limit grain yields. When soil moisture was limiting, grain yields were not af-
fected by increased seeding rates, whereas at high yields, grain yields varied in re-
sponse to changes in seeding rate, with grain yields more responsive to seeding
rates in 18-cm row spacings than in 76-cm row spacings. Therefore, environment
50 FRANCIS J. PIERCE AND PETER NOWAK

and row spacing interact with seeding rate in determining the success of variable
seeding rate for soybeans. Pioneer Hi-Bred (1997) reported that narrow rows for
corn production increased yields but increased plant populations were not neces-
sary for increased yields at narrow row spacings. However, increasing plant pop-
ulations beyond 3 1,OOO plants per acre continued to increase yields in narrow rows
but not in wide rows. Sojka ef al. (1988) identified the manipulation of row spac-
ing as the most practical means of optimizing canopy geometry and this was sup-
ported by the data of Porter et al. (1997) for corn.
Theoretically, opportunities for precision crop management appear to exist, but
early field evidence does not support it. To some extent, a lack of response may re-
sult from plant breeding programs developing cultivars that tend to perform well
over a range of environmental conditions rather than to optimize for specific en-
vironments. Comparison of cultivars under varied management developed under
this scenario is a self-fulfilling prophecy in the sense that cultivars created to per-
form on the average will on average outperform those that do not. Whether breed-
ing will focus on improvements in traits other than yields or on yielding ability
when nutrients and water are suboptimal or continue to focus on yield ability in
normal or highly favored environments is under discussion (Duvick, 1996). Re-
gardless, a glance at the plethora of yield maps revealing spatial variability in crop
yield is sufficient to raise expectations about what might have happened if a dif-
ferent cultivar had been planted or something different had been done with respect
to planting geometry. Certainly, something as simple as varying planting depth
based on soil moisture in the seed zone (Price and Gaultney, 1993; Weatherly and
Bowers, 1997) should find some utility in production agriculture. We will not
know what precision crop management can or will bring to production agriculture
until it becomes a research priority in the crop science community.

d. Precision Water Management


In most crops, growth can proceed unimpaired and crop yield can be maximized
only when the soil moisture potential remains high (and water remains readily
available) continuously throughout the growing season.
-Hillel (1990)
Water is critical to crop productivity since crop yields generally increase lin-
early with water transpired by a crop (Howell, 1990).Excess water (waterlogging)
can induce nutrient and aeration stresses and encourage pests that reduce yield and
quality (Wesseling, 1974; USDA-ERS, 1987). Water management is also critical
to water quality because techniques to optimize water relations for plants can also
impact fate and transport of pollutants to surface water and groundwater. Natural-
ly, the adequacy of water for plant growth is primarily related to the amount, fre-
quency, and distribution of rainfall, soil properties as they affect processes that reg-
ulate soil water availability to plants, and landscape properties that regulate the
ASPECTS OF PRECISION AGRICULTURE 51

hydrologic cycle within a watershed. Three approaches to precision water man-


agement are therefore apparent: (i) variable rate irrigation, (ii) matching agro-
nomic inputs to water availability defined by soil and/or landscape properties, and
(iii) drainage.
i. VARIABLERATEIRRIGATION.Hillel (1990) defines a well-managed irriga-
tion system as one that optimizes the spatial and temporal distribution of water so
as to promote crop growth and yield and to enhance the economic efficiency of
crop production (maximum net return). He further states that since the physical
circumstances and the socioeconomic conditions for irrigation are site specific
(and often season specific) in each case, there can be no single solution to the prob-
lem of how best to develop and manage an irrigation project. Hillel unknowingly
defined the rationale for variable rate irrigation management.
Considerable progress has been made with variable rate irrigation systems pri-
marily with sprinkler irrigation provided by center-pivot and linear-move ma-
chines (Camp and Sadler, 1994; Evans et ul., 1996; King et ul., 1995, 1996; Mc-
Cann and Stark, 1993).These site-specific irrigation systems require high spatial
resolution (currently 10-30 m) achieved by adding more discrete control between
contiguous elements of the machine, all at higher costs than those of current sys-
tems (Sadler et al., 1998).Variable irrigation is coupled with precision nutrient and
pest management via chemigation, in part because variable irrigation facilitates in-
creased management precision in space and time (King et al., 1995) and in part
because it may not be economically feasible to site-specifically manage only for
water (Evans et al., 1996).The uniformity of chemical application depends on the
uniformity of water application (King et al., 1995), requiring injection equipment
that can vary the amount of chemical injected into the boom in proportion to the
flow rate of water in order to achieve the desired chemical application rate (Sadler
et al., 1996).
Success of precision irrigation management has been achieved with regard to
application control (Camp etal., 1996;Wall et al., 1996).The key to the agronomic
success of precision irrigation management depends to a large extent on how well
the water needs of the soil-plant system can be measured or predicted and the ac-
curacy of water application (and agrichemical) prescriptions. The value of preci-
sion irrigation management depends on whether increased profits and the reduc-
tion in pollutants more than offset the cost of increased resolution needed in
irrigation systems to apply irrigation and chemigate site specifically. Evans et al.
(1996) concluded that the hardware, software, and communication systems to de-
liver a prescription work well but a major limitation lies in the ability to interpret
spatially variable data and develop rational and coherent site-specific prescrip-
tions. They further conclude that
Due to the random variability in water application distributions due to wind,
start-stop operations of the self-propelled machines, and sprinkler pattern vari-
52 FRANCIS J. PIERCE AND PETER NOWAK

ations combined with the low cost of water and N fertilizers, it is probably not
economically feasible to site-specifically manage only for water and/or nitro-
gen.
King et al. (1995) agree that the lack of a service infrastructure to generate and
deliver the maps needed to control and manage the irrigation system throughout
the season remains a major limitation to variable water application. They identi-
fied the lack of variable rate sprinklers as the other major limitation, although they
overcame this limitation to some extent by the use of multiple sprinklers and siz-
ing the sprinkler nozzles to provide a stepwise variable application rate. Because
of the large temporal variability in water availability, sensing plant stresses to de-
termine water application rates may be useful (Stone et al., 1996), but its efficacy
is not fully determined. Like other factors, variable imgation management requires
sufficient knowledge about the spatial and temporal variability of the factors that
regulate water availability to plants and water use in plants in order to achieve pre-
cision management. Thus, its potential will vary by soils, landscape, and climate.
ii. SOIL-LANDSCAPE WATER MANAGEMENT. The potential for precision man-
agement of agronomic inputs increases with spatial variability in water availabil-
ity within a field. Differences in water availability within a field are governed by
(i) the Occurrence of dissimilar soil types; (ii) the presence of soil degradation
processes (e.g., erosion, compaction, and salinization); and (iii) variation in land-
scapes. The evidence for spatial variation in water availability is clear. Hanna et
al. (1982) reported that north-facing slopes had 20% more available water in soils
than south-facing slopes throughout the year, whereas soils on east-facing slopes
were the driest. Crop yields are often highest in the lower slope positions where
soil water and nutrient contents are higher (Fiez et al., 1994a,b; Halvorson and
Doll, 1991; Jones et al., 1989; Mulla et al., 1992; Spomer and Piest, 1982). Erod-
ed soils often have lower infiltration rates and lower available water than their
noneroded counterparts (Daniels et al., 1985; Langdale et al., 1979). Some por-
tion of landscape variability can be attributed to the variation of soil properties
with landscape position (Brubaker et al., 1993; Khakural et al., 1996; Kreznor et
al., 1989; Walker et al., 1968), whereas some is attributable to redistribution of
water within a landscape due to either runoff or subsurface horizontal flow of wa-
ter (Miller et al., 1988; Stone et al., 1985). Compacted soils reduce infiltration or
restrict plant roots, thereby limiting water availability to plants (Lindstrom and
Voorhees, 1994). Areas of high salinity are known to reduce yields (McKenzie et
al., 1983). Nonuniformity is easily demonstrated in dryland areas of the United
States where crops generally use all the water that is available each year and stored
soil water is an essential source of water for crop production (Hanna et al., 1982).
Additionally, topographic trends in soil water storage or crop growth can be im-
portant even where it is not expected, such as the sand-plain landscape in central
Minnesota (Tomer et al., 1995).
ASPECTS OF PRECISION AGRICULTURE 53

Knowledge of the spatial distribution of water availability can be used as a ba-


sis for site-specific input recommendations. There are three approaches for map-
ping soil water variability (Bell et al., 1995): (i) county soil surveys; (ii) interpo-
lation of a network, usually a grid, of point samples to estimate spatial distribution
of soil properties or water content; and (iii) soil-landscape models to estimate spa-
tial patterns of soil water availability. The internal variability of soil map unit de-
lineations in Order II county soil surveys may limit their use in site-specific man-
agement (Kellogg, 1961; Mausbach et al., 1993), although where soil map units
are dissimilar, map units may correspond to crop yield variability. Interpolation
techniques are used extensively in precision agriculture and their use and lim-
itations have been discussed previously. The presence of small-scale spatial
variability in soil physical properties (Peck, 1983) and the high cost of network
sampling may limit its use in mapping water availability. Site-specific soil water
monitoring is used to some extent as a basis for variable rate irrigation (Evans er
al., 1996) and in landscape studies (Khakural et al., 1996). Statistical models of
soil-landscape relationships offer opportunities to map spatial patterns of soil
properties where relief or some landscape attribute is a primary factor contributing
to soil variability (Bell et al., 1992). Soil-landscape models are important because
terrain modifiesthe distributionof hydrologic and erosional processes (i.e., soil wa-
ter content, runoff, and sedimentation)and soil temperature in fields (Moore et ul.,
1993), all important in regulating crop productivity and off-site movement of agri-
chemicals. The depiction of the spatial variability of topography with a regular grid
of elevation observations is referred to as a digital terrain model (DTM) when at-
tributes of a landscape are of interest and a digital elevation model (DEM) when
merely relief is represented (Weibel and Heller, 1991). A DTM allows the estima-
tion of derivatives of elevation including slope, curvature, aspect, catchment area,
and surface drainage proximity variables that correlate to soil and land qualities
(Bell et al., 1992, 1994; Moore et al., 1993; Odeh et al., 1995). A comprehensive
review of relevant techniques and applicationsof DTM is presented by Weibel and
Heller (1991). The value of DTM is that it increases the resolution of soil maps for
use in site-specific management and in environmental modeling by using terrain
attributes to spatially distribute estimated soil attribute data (Moore et al., 1993).
Therefore, terrain modeling efforts have focused on its application to soil survey
to model and depict the spatial variability of soil horizons in reference to the topo-
graphic surface (Bell et al., 1995) and spatial application of simulation models to
evaluate current and potential management practices regarding their effects on crop
production and the environment in space and time (Verhagen et al., 1995a,b). The
extent of use of soil-landscape models is currently limited in field applications of
precision agriculture. However, high-resolution DEMs can easily be created using
DGPs and laser-based systems with high vertical accuracies (Clark, 1996; Lange,
1996). As elevation maps become available, soil-landscape modeling techniques
such as DTM will be increasingly used in precision agriculture.
54 FRANCIS J. PIERCE AND PETER NOWAK

iii. DRAINAGE.Poor drainage is often cited by farmers as a source of yield


variability within fields. Many options for drainage currently exist and can be ap-
plied site specifically.Therefore, there is little need to design site-specificdrainage
practices. The decision to install drainage is economically, not technically, limit-
ed. Regardless of scale, the decision to drain hinges on the expectation of returns
on investment that exceed costs of installation. Site specifically, the cost of drain-
ing portions of fields or small isolated areas may be higher because costs to con-
nect to drain outlets are proportional to the distance to the outlet. Yield mapping
and GIS have made the cost/benefit calculation easier because the yield depres-
sion due to poor drainage can be accurately assessed if sufficient years are includ-
ed in the calculation. Drainage, therefore, is a site-specific, economic decision
based on the conditions at each site and cannot be generalized.

3. Evaluation of Precision Agriculture

We have probed the technological capabilities and agronomic feasibility of pre-


cision agriculture, i.e., the technologies and techniques for assessing and manag-
ing spatial and temporal variation. Essentially, we have argued that initial forms
of a precision agriculture system are technologically feasible and based on credi-
ble agronomic principles. However, being technologically possible while being
based on sound scientific principles does not necessarily establish utility or value
in the process. An evaluation of precision agriculture is also required. Three im-
portant evaluation issues surrounding precision agriculture remain unresolved:
economics, environment, and technology transfer. The economic evaluation fo-
cuses on whether the documented agronomic benefits-translated into value
through market mechanisms-exceed the technological and service costs. Envi-
ronmental evaluation focuses on whether precision agriculture can improve soil,
water, and the general ecological sustainability of our agricultural systems. Final-
ly, and perhaps most important, is the question of whether this bundle of enabling
technologies and agronomic principles will work on individual farms. Being tech-
nologically feasible and at least economically neutral are necessary conditions but,
as will be shown, may not be sufficient conditions for transfer to farms. Examin-
ing precision agriculture from the perspective of the technology transfer provides
a context for the ongoing debate over the scale neutrality of this technology.

a. Economics
The most important fact regarding the analysis of the profitability of precision
agriculture is that the value comes from the application of the data and not from
the use of the technology. This can be contrasted with traditional agricultural in-
novations in which the value is derived from the use of the new technology, e.g.,
new seed genetics that increase yield or a new herbicide that reduces yield loss.
However, the enabling technologies discussed earlier only generate data that have
ASPECTS OF PRECISION AGRICULTURE 55

to be analyzed into practical information (Boehjle, 1994) so that it can influence


management decisions. If any value is to be derived from precision agriculture, it
will come from the resulting management decisions and not through the use of the
enabling technologies. The data have no value. It is only through the interpreta-
tion and application of data that value may be derived (Lowenberg-DeBoer and
Boehjle, 1996).This is a critical distinction that needs to be emphasized and reem-
phasized because it dictates two postulates that should drive any analysis of the
economics of precision agriculture.
First, a theoretical or modeled analysis of the economics of precision agricul-
ture will not necessarily have the same result as an actual application on a farm.
One can model the costs of the enabling technologies (Kohls, 1996) and, based on
agronomic principles, determine the scope and extent of differences in production
efficiency (i.e., increases in yield per unit of input). However, an actual applica-
tion on a farm may have different results because of variation in the managerial
capacity of the operator relative to responding to temporal events such as climate
and pest or market cycles. The profitability will differ if a piece of equipment was
not properly calibrated or erroneous agronomic decision rules were used. In sum-
mary,real-world profitability of precision agriculture is highly dependent on the
human capability to manage dynamic forms of spatial and temporal variability.
Second, the potential profitability of precision agriculture is directly related to
the nature and extent of variability in the biophysical setting in which it is applied.
If this biophysical setting was homogeneous, then there would be no difference in
profitability between a precision and a conventional (“one rate fits all”) agricul-
tural system. As the heterogeneity in this biophysical system increases, especially
in those salient situations or processes in which valid agronomic principles exist,
then the potential for profitability will increase. Of course, at some point in this
implied linear relationship the extent of heterogeneity will overwhelm the current
capability of the enabling technologies and the agronomic principles. Nonetheless,
the potential profitability of precision agriculture is strongly correlated with the
biophysical heterogeneity of the setting to which it is applied.
At issue is the extent that these two postulates have influenced research on the
profitability of precision agriculture. Evidence of the economics driving the adop-
tion of precision agriculture is fragmentary and incomplete relative to these pos-
tulatesfor several reasons. First, as has been noted (Lowenberg-DeBoer and Swin-
ton, 1997; Nowak, 1997), there is no precision agriculture system. Rather, there
are different clusters of technologies and tools that may be adopted in different se-
quences and combinations,each of which may be used with varying degrees of ef-
ficiency under different cropping system conditions. Moreover, the impacts of pre-
cision agriculture applications may be masked by the interaction with other new
technologies or techniques simultaneouslyintroduced into the operation. New her-
bicides, crop genetics, or tillage tools may influence the overall profitability of a
production system. In summary, it is difficult to assess the profitability of a system
56 FRANCIS J. PIERCE AND PETER NOWAK

when there is little consistency in the nature of the system between comparative
sites.
At the other end of the spectrum is what may be called a partial analysis of the
profitability of precision agriculture. To date, the profitability of precision agri-
culture has only been addressed by a limited number of case studies, focus groups,
antidotal stories, or modeling efforts (Denton, 1996; Holmes, 1993; Hornbaker,
1996; Olson, 1995; Schnitkey er al., 1996; Swinton and Ahmad, 1996). Much of
this effort is focused on examining the profitability of a specific aspect of preci-
sion agriculturefor a specific producer or a specific cropping practice such as fer-
tility. No research studies have examined the economics of precision agriculture
for an agroecological region, the type of production system, or across all dimen-
sions of a production system. Many different factors and methods may be em-
ployed in an analysis of the profitability of precision agriculture (Fairchild, 1993).
Lowenberg-DeBoer and Swinton (1997) analyzed many studies that used partial
budget analysis but were characterized by little consistency in exactly what costs
were considered. They suggest a broader investment analysis to capture the flow
of costs and benefits across time. A related problem is the relative ease of captur-
ing costs (Kohls, 1996), whereas benefits are more difficult to conceptualize and
measure. Besides yield differences or reduction in input costs (i.e., efficiency), pre-
cision agriculture may also have benefits with regard to managing differentiated
products, food safety, increasing the span of control, and environmentalprotection
(Lowenberg-DeBoer and Boehjle, 1996; Nowak,1997).
Future research on the economics and adoption of precision agriculture need to
carefully consider the earlier postulates on how precision agriculture differs from
traditional agricultural technologies. While there is consensus that profitability
will dictate the adoption rate of precision agriculture technologies, care needs to
be exercised in the dominanceattributed to economic rationality in these decisions.
This comment is based on the apparent lack of economic rationality driving many
of the adoption decisions reported in the media and case studies. That is, the adop-
tion decision does not appear to be driven by proven short-term profits based on a
partial budget analysis or long-term gains substantiated by an investment analy-
sis. Rather, one reads in this collection of adoption stones a strong underlying be-
lief that the information-especially that associated with spatially referenced yield
data-will be valuable even if that value does not manifest itself in the short term.
The spatial and temporal dimensions of precision agriculture may represent a new
type of information whose value has not yet been established under traditional ac-
counting procedures or existing market mechanisms. Profitability remains impor-
tant, but current adopters of precision agriculture also report a belief that the in-
formation collected under this method will somehow, sometime pay for itself. For
example, in addition to the potential to enhance production efficiency, this new
type of information has been associated with on-farm experimentation, thus re-
ducing dependence on external parties (Fortin and Pierce, 1998). It can also pro-
ASPECTS OF PRECISION AGRICULTURE 57

vide value to the farmer in evaluating the worth of products and services offered
by vendors. Furthermore, aggregated data from local firms offering mapping or
analysis services may acquire additional value for private-sector vendors who de-
velop marketing or promotional strategies based on product efficacy across a lo-
cal area or region (Nowak, 1997). Finally, there is the potential to quantify envi-
ronmental benefits for either self-satisfaction or regulatory avoidance (Larson et
al., 1997; NRC, 1997).Consequently, rather than strict economic rationality, many
current adoption decisions appear to be driven by a future and unknown potential
value where the adopter is waiting for the science to “catch up” with technologi-
cal applications.
A more traditional economics-oriented analysis of the potential profitability of
precision agriculture is found in the NRC (1997) report. This discussion, howev-
er, results in three paradoxical conclusions. First, the potential profitability of pre-
cision agriculture is directly related to the extent of natural and anthropogenic
sources of manageable variance. Producers who have effectively used analytical
tools (e.g., soil testing and scouting) to manage these sources in the past actually
have less potential for profit versus producers who have used a “one-size-fits-all”
farming approach. At the same time, the prediction of the traditional adoption mod-
el (Rogers, 1995)that the more sophisticatedor better managers will be among the
first to adopt appears to be holding relative to precision agriculture. This leads us
to the conclusion that those who will benefit the least economically are among the
first to use precision agriculture.
The second paradox is related to the sources of variation and the potential prof-
it to be derived from the process of applying precision agricultural technologies.
While indicators of the adoption of precision agriculture appear to be following
the traditional logistic rate of adoption (Rogers, 1995), the greatest sources of po-
tential profit have yet to be developed. Potential profits from managing nutrients
that are relatively stable in space and time (P and K) are significantly less than prof-
its from managing weeds, insects, disease, water, and genetics across space and
time. These latter factors are much more spatially and temporally diverse, repre-
sent a greater proportion of production expenses, and generally have a greater im-
pact on production outcomes. They will also require further development of the
enabling technologies and agronomic decision rules if these larger profits are to be
realized. Consequently, the adoption and diffusion of precision agriculture is con-
tinuing at expected rates, whereas the greatest benefits from this technology are
yet to be realized.
One might paraphrase the English poet John Donne by stating that “no man is
an island onto himself’ when it comes to the process of managing precision agri-
culture for profit. While we have constantly alluded to the farmer or producer
adopting and managing these enabling technologies, the simple fact of the matter
is that the outcome of this process is highly dependent on the support of suppliers
of products and services. This infrastructure, a concept discussed more fully later,
58 FRANCIS J. PIERCE AND PETER NOWAK

has a major influence on the ability of any producer to achieve profits with preci-
sion agriculture. Thus, the third paradox surrounding the economics of precision
agriculture is the fact that the extent of profitability for any producer will be strong-
ly influenced by the quality of the local infrastructure. Precision agriculture is a
“team” activity, and although there are a few individuals who can make it on their
own, the vast majority of farmers will require quality support products and ser-
vices if they are to achieve a profit.

b. Environment
In our opinion, one of the greatest constraints to managing for improving water
quality is the inability of agricultural producers to control inputs in ways that
account for the positional and temporal variability in growing conditions across
a field.
-Evans et al. ( 1996)
Potential improvements in environmental quality are often cited as a reason
for using precision agriculture (NRC, 1997). Reduced agrochemical use, higher
nutrient use efficiencies, increased efficacy of managed inputs, and increased pro-
tection of soils from degradation (erosion) are frequently cited as potential bene-
fits to the environment (Engel and Gaultney, 1990; Larson et al., 1997). Some
pose precision agriculture as a “win-win for the environment” (Farm Industry
News, 1994). While the impacts of precision agriculture on the environment are
assumed positive, proof that it is even benign in its environmental impacts is gen-
erally undocumented. Studies evaluating the environmental benefits of precision
agriculture are limited (Khakural ef al., 1995; Mulla et al., 1996; Verhagen et al.,
1995a). Negative impacts of precision agriculture could derive from increased ag-
gression by farmers on site-specific potentials or problems (Pierce, I997b). Fur-
thermore, many factors may limit reductions in chemical applications in some sit-
uations and encourage increased use in others (NRC, 1997). Some doubt that
precision agriculture can materially improve the environmental performance of
agricultural production systems because current systems are fundamentally
flawed compared to natural ecosystems and precision agriculture cannot fix these
flaws. Groffman ( 1997) argues that precision agriculture techniques do not ad-
dress many of the key factors that cause poor environmental performance in agri-
cultural systems, that agricultural inputs are high relative to ecological processes
we hope to manage, and that these processes vary on scales that are incongruent
with precision management techniques. Specifically, he discusses four key fac-
tors that constrain the environmental performance of agricultural production sys-
tems: (i) temporal discontinuities in nutrient cycling processes (a non-crop peri-
od that provides no plant nutrient sink to prevent losses); (ii) high levels of soil
disturbance (mainly through tillage) that exacerbates temporal discontinuities in
nutrient cycling processes, creates small-scale spatial discontinuities in this
ASPECTS OF PRECISION AGRICULTURE 59

process, and increases the susceptibility of agricultural systems to erosion; (iii)


high levels of nutrient enrichment that foster high losses unless fertilizer use is re-
duced dramatically; and (iv) lack of resistance to the disturbance of extreme cli-
matic events (i.e., high rainfall) that cause a major portion of nutrient and sedi-
ment losses. The uncertainty about environmental benefits of precision
agriculture is underscored by the fact that environmental improvements by them-
selves offer little incentive for farmers to adopt precision agriculture and will be
an incentive only where producers bear at least a share of the cost of agricultural
pollution (NRC, 1997).
The notion that agriculture is inherently leaky is reason to pursue the potential
of new technologies such as precision agriculture. That soils vary considerably in
their potential to pollute the environment further supports this view. For example,
soil erosion (Larson er al., 1983)and nitrate leaching (Mueller er al., 1995)are ma-
jor problems on only a portion of cropland in the United States and can vary with-
in fields (Larson et al., 1997). We started this discussion with the notion that the
temporal component of spatial variability is what makes precision agriculture dif-
ficult but perhaps more rewarding. While it is probably not possible for precision
agriculture to make agricultural production systems as tightly coupled to ecolog-
ical processes as Groffman (1997) would like, there are many opportunities to re-
duce inputs, increase use efficiencies, and protect the soil using precision agricul-
ture technologies than there are without using them. We will briefly discuss the
potential opportunities for precision agriculture to improve environmental quali-
ty. However, keep in mind that evaluating the environmental performance of pre-
cision agriculture is problematic since all aspects of precision agriculture to date
are merely components of a system rather than a system in itself (i.e., no precision
agricultural systems exist).
Pollution of the environment by agriculture requires the presence of a pollutant
in a form that is suitable for transport in air or water by suspension, volatilization,
or dissolution. Precision agriculture can be of benefit to the environment by re-
ducing the application of a given input, affecting its mobility or persistence, and/
or regulating the mechanisms responsible for its transport. Farmers must manipu-
late these aspects not discretely but over the continuum of space and time repre-
sented by the nature of inputs (type, quantity, volatility, sorptivity, solubility, and
persistence) and the heterogeneity of the biophysical environment where inputs
are introduced. Environmental benefits of precision agriculture could include eco-
logically based management practices such as the introduction of refugia within
fields or landscapes to preserve or enhance biological and genetic diversity (Fleis-
cher et al., 1997) and precision management of the non-crop period (Pierce and
Lal, 1991).
The environmental impacts of precision agriculture will be difficult and costly
to quantify, particularly as the temporal component of variability increases. What
follows is a brief listing of the areas in which potential environmental benefits of
60 FRANCIS J. PIERCE AND PETER NOWAK

precision agricultureexist (We recognize that these are not well documented in the
literature):
Reduction in nutrient inputs where nutrient levels or supply capacity are suffi-
cient to meet the nutritional requirements of crops: This may have short-term
benefits in areas in which residual nitrates can be utilized by the crop and there-
by prevented from leaching (Hergert er af., 1996) but may not have short-term
benefits, as in the case of P, in which it may take decades for soil test levels to
be depleted from high levels (McCollum, 1991). Addition of nutrients where
needed to meet crop nutritional requirements should have minimal impact on the
environment.Thus, redistribution of nutrients through precision agriculture may
or may not decrease total fertilizer use but may have environmental benefits
within fields.
Reduction in pesticide inputs through variable rate applications:This can be par-
ticularly useful in patch spraying where significant portions of a field may not
receive any pesticide treatment (Johnson et af., 1997).
Reduction in irrigation water inputs in areas subject to leaching using variable
rate imgation.
Minimizing or avoiding nutrient and pesticide additions where the potential for
significant losses exist: This can be accomplished by varying nutrient or pesti-
cide type, formulation, and rate according to soil conditions for erosion, leach-
ing, runoff, and volatilization. For example, Sharpley (1997) recommends the
use of environmental soil tests for P and the delineation of transport zones with-
in a field to minimize the losses of P to surface waters.
Increased erosion control or a reduction in runoff achieved through site-specif-
ic tillage and residue management: Different parts of a field erode at different
rates, making precision tillage and residue management very desirable (Voor-
hees ef al., 1993). Pierce and Gilliland (1997) suggest that precision agriculture
provides the necessary level of design and control needed to achieve soil quali-
ty control on the land. It is quite remarkable that an analysis of yield maps has
not been performed to demonstrate the effects of erosion on productivity, there-
by quantifying the importance of erosion control on the farm.
Management of field and landscape buffer zones not possible without precision
management systems. Because of spatially variable control, buffer zones and
refugia can be managed without interruption of farming operations creating pos-
sibilitiesfor increased biological diversity and interception of pollutant transport
to surface water.
Although precision agriculture is intuitively appealing, environmental benefits
are not necessarily forthcoming. However, there is equally little evidence to sug-
gest that precision agriculture will increase environmental pollution or that it can-
not, as Groffman (1997) suggests, address the fundamental environmental prob-
lems facing agriculture.The best interest of agriculture is served by documenting
ASPECTS OF PRECISION AGRICULTURE 61

the environmental performance of precision agriculture over the space-time con-


tinuum.

c. Technology Transfer
The third critical dimension of evaluating precision agriculture involves exam-
ining the technology transfer process. Enabling technologies can make precision
agriculture feasible, agronomic principles or decision rules can make it applica-
ble, and enhanced production efficiency or other forms of value can make it prof-
itable. However, will precision agriculture be used on a farm? The reality check
for this scientifically feasible and potentially economically viable bundle of tech-
nology comes from an assessment of what must happen on a farm to make it work.
We know that producers have three basic questions relative to the adoption of pre-
cision agriculture: Will it work? Will it pay? and Can I make it work in my oper-
ation? While basic and applied research may answer the first two questions, the
third question is highly dependent on farmers and the context in which they oper-
ate. Research of a different sort is required to answer this question-research that
has been subnamed under the label of technology transfer.
The term technology transfer could imply that precision agriculture occurs when
individuals or firms simply acquire and use the enabling technologies (i.e., it is
transferred to them). While precision agriculture does involve the application of
enabling technologies and agronomic principles to manage spatial and temporal
variability, the key term is manage. Much of the attention in what is called tech-
nology transfer has focused on how to communicate with the farmer (Crookston,
1996; Holt and Sonka, 1995; Peterson and Beck, 1997). However, communicat-
ing the answers to the three basic questions of a farmer listed previously is differ-
ent than the substantive nature of those answers. If farmers are going to manage
precision agriculture, then we need to understand what will influence this man-
agement process. It is here, and not in communicationtechniques, that we will be-
gin to understand what drives the technology transfer process. Three factors will
influence the transfer or diffusion of precision agriculture among a population of
farmers: distributions of human capital, spatial locations and the nature of infra-
structure support, and the compatibility of the enabling technologies to the needs
and capabilities of producers.
According to Bouma (1997), farmers make strategic, tactical, and operational
decisions relative to agriculture in general and precision agriculture in particular.
Strategic decisions surround the nature of the farm such as the use of crop rota-
tions, tactical decisions focus on things such as crop varieties, and operational de-
cisions are those focused on mounts and timing of inputs for the selected crop va-
rieties. Bouma (1997) observes that most precision agriculture research has
focused on improving the efficiency of operational decisions. However, the deci-
sion to adopt precision agriculture clearly occurs at the strategic level. Therefore,
technology transfer is the process of getting farmers to make a strategic decision
62 FRANCIS J. PIERCE AND PETER NOWAK

on how operational decisions will be made in the future. How do we explain why
some farmers make this decision whereas others remain skeptical, uninformed, or
negative regarding precision agriculture?
On an individual level, perhaps the best predictor of the adoption and diffusion
of precision agriculture will be based on the classic adoption model (Rogers,
1995), which places a heavy emphasis on the social psychological or human cap-
ital attributes of the potential adopter. Others have already hypothesized that the
adoption of precision agriculture will follow the pattern predicted by this model
(Lowenberg-DeBoer, 1997; NRC, 1997). Individuals who have greater risk pro-
pensity and who are younger, better educated, able to manage transition costs, and
are well integrated into diverse information networks will be among the first to
adopt precision agriculture. Individuals who are among the first to adopt are called
“innovators” in the model and comprise approximately 2.5% of the population.
These innovators then influence the early adopters in local areas or neighborhoods
who comprise the next 13.5%who adopt the innovation. They are followed by the
early majority and late majority, each comprising 34% of the population. The fi-
nal group is composed of those who are either unable or unwilling to adopt for a
variety of reasons. This “trickle down” process through social contagion and mar-
ket processes has been well documented (Rogers, 1995).
However, an individual’s social-psychological, economic, and demographic
profile may be necessary, but it is not sufficient for the adoption of precision agri-
culture. Access to resources and support through local infrastructure is also criti-
cal for adoption. That is, one may be an innovator, but without access to dealer-
ships, consultants, experienced farmers, manufacturing representatives, or other
experts, adoption of precision agriculture is more difficult. A parallel can be found
in the computer industry. There is a well-developed computer industry infrastruc-
ture in the “Silicon Valley” of California.The entrepreneur in this area is support-
ed by the levels of innovation, adaptation, and competition for success, whereas
entrepreneurs in other regions are penalized because they do not have access to
similar supporting resources. It is relatively easy for the hardware or software de-
signer to find supporting parts, services, and ideas that may not be available to the
entrepreneur in the midcontinent or other regions. The same situation also applies
to agriculture in general and precision agriculture in particular. Since private and
public supporting resources are not equally distributed across the agricultural pro-
duction landscape, some producers are going to have a greater opportunity to adopt
precision agriculture than others regardless of individual managerial or economic
attributes. Related to this notion of a lumpy distribution of supporting resources is
the relative role of the private and public sectors. The role of infrastructure in dif-
fusing precision agriculture is contrary to many claims of the emerging informa-
tion age. Here, information technologies are supposed to create equal access to
ideas, data, and management information. This “leveling of the playing field” due
to information technologies, however, is not occurring with precision agriculture,
ASPECTS OF PRECISION AGRICULTURE 63

due largely to the complexity and fragmentation of current tools and analytical
processes. Making precision agriculture work on a farm is a very difficult process
involving contradictory agronomic recommendations, incompatible hardware,
and high learning costs with few specific objectives (other than profitability). All
this creates a situation in which contracted expertise, products, and services from
off the farm become the only viable strategy. This situation will change as more
protocols, standards, and a uniform set of expectations are developed, at which
point precision agriculture will help agriculture move into the information age.
What is the importance of the private sector relative to the public sector in in-
fluencing the strategic decisions of farmers? As Wolf (1998) notes,
After more than 100 years of public-sector leadership in agriculture research
and extension in the United States, we appear to be entering a new era in which
private and corporate interests have significantly increased authority and re-
sponsibility as information providers.
In essence, the private sector largely dictates the production decisions of com-
mercial farmers in the United States today. Since precision agriculture is being
largely designed for commercial applications, we would expect the private sector
to have an inordinate influence on current technology transfer efforts for precision
agriculture, affecting the geographical areas where it is available, the extent to
which the various products and services are accessible to the producers in these
areas, and the marketing strategies used to promote precision agriculture. Dealer-
ships, consultants, or manufacturing representatives,like the customers they serve,
can be aggressive or conservative relative to new technologies. Furthermore, the
geographic concentration of customers largely dictates the location of these sup-
porting facilities due to basic market mechanisms.Agricultural technologies, pre-
cision or otherwise, are designed based on potential market demand and techno-
logical feasibility. Factors such as the number of potential customers, their ability
to invest in new technology, and the geographic concentration of these customers
all help to determine the spatial market demand for precision agriculture. Public-
sector organizations, such as the land grant university or the USDA, will focus
their precision agriculture efforts on problem solving, efficacy testing, and basic
research. In essence, however, it needs to be emphasized that most of the technol-
ogy transfer surrounding precision agriculture will be market driven from the pri-
vate sector, commodity specific, and concentrated in certain agricultural areas.
The final point about technology transfer concerns the compatibility of the en-
abling technologies to the needs and capabilities of the farmers. Much could be
said about this issue, but attention has focused on whether precision agriculture is
a scale-neutral technology.A pure economics approach to this issue would address
whether there are equivalent proportionate investments across scales of opera-
tions. This approach, however, ignores several spatial dimensions associated with
access to this technology. First, precision agriculture technologies have not been
64 FRANCIS J. PIERCE AND PETER NOWAK

developed and adapted to work with all agricultural commodities. There was a
clear sequence of developmentfrom certain cash grains to some horticulture crops
to crops such as cotton. Producers of some agricultural commodities,regardless of
scale of operations, cannot adopt this technology because it has not been devel-
oped for their commodity. Second, as discussed earlier, the support infrastructure
has a lumpy spatial distribution. Producers in some areas will either not be able to
adopt or have a very difficult time adopting because of the lack of these support-
ing resources. Finally, farmers are not equivalent relative to issues such as com-
puter literacy, technical competence, or abstract reasoning. To the extent that cur-
rent forms of precision agriculture call for these skills, it is not compatible with a
certain proportion of these farmers. Consequently, while the scale neutrality of pre-
cision agriculture has been a contentious issue, understanding what influences the
potential management of precision agriculture provides more insight than seman-
tical debates. Precision agriculture is not scale neutral for the reasons discussed as
part of the technology transfer process. Nor is there any inherent reason why it
should be other than a social value related to equity. Understanding why this is the
case should move the debate out of the normative and political arena and allow fu-
ture research to focus on how to accelerate technology transfer for targeted regions
of the country.
What happens as new products or management techniques are designed specif-
ically to work under precision agriculture?Consider the commercial release of the
Windows 98 operating system. There may be little interest in moving to this new
operating system as users perceive little value in that decision. The utility of a new
operating system will emerge as developers create useful products that will only
operate efficiently under Windows 98. The same was true of the transition from
Windows 3.1 1 to Windows 95. An analogous situation exists relative to precision
agriculture, a new operating system for crop production. Precision agriculture pre-
sents a new set of diagnostic, analytical, and management tools oriented toward
spatial and temporal variation that must produce utility with old products and pro-
duction models. Much of the current agricultural machinery, nutrients, pesticides,
and genetics were designed to work with the old one-size-fits-all agricultural sys-
tem that has evolved since the 1950s (Nowak, 1997). Should new products and
techniques specifically designed to work in the new operating system of precision
agriculture emerge, the paradigm for technology transfer changes dramatically to
one in which to reap the benefits of new products and practices farmers must be
equipped for precision agriculture. Consider, for example, current genotypes that
optimize performance across a wide range of microenvironmentsmay not be able
to compete against a suite of new genotypesthat maximize performancein selected
microenvironments. It is conceivable that a series of these new genotypes devel-
oped just for precision applications will significantly outperform the old one-size-
fits-all genotype currently available. Few have yet to realize how emerging bio-
technology applications will also support precision agriculture and vise versa.
ASPECTS OF PRECISION AGRICULTURE 65

Imagine nutrients whose formulation is designed for microenvironmentsbased on


reducing mobility while also being in synchronization with the needs of the plant
across the growth cycle. Similar “designer” approaches can be developed relative
to pesticides and water management. In short, we have been evaluating the utility
of a new production system under the playing rules of the old system. The tech-
nology transfer issues associated with the managerial capability of the operator,
the spatial distribution of infrastructure,and the compatibility of the technology to
individual farms will change radically as precision agriculture continues to devel-
op. Anticipating these changes is important for evaluation. It is also important for
understanding how the assessment and management of variation in precision agri-
culture must be performed. Researchers need to examine current developments.
However, we also need to give attention to the “what could be” futures for preci-
sion agriculture.

III. CONCLUSIONS

The value of the emerging system of precision agriculture will ultimately be


measured by its success in managing the space-time continuum of all aspects of
crop production. This, in turn, will have implications for environmentalprotection
and farm profitability. Early successes in precision management have come where
spatial dependence is moderate to high and temporal dependence is low, e.g., cer-
tain aspects of soil fertility and weed patch management. We believe that the long-
term application of precision management where temporal dependence is low will
reduce spatial dependenceand increase predictability. This should make precision
management easier and more efficient as a system but potentially less profitable
with respect to a given component. For example, through precision management,
soil test P levels within a field will evolve to the maintenance plateau levels, there-
by linking P fertilizer rates to crop removal rates estimated from yield maps. For
spatially dependent entities, high temporal dependence makes precision manage-
ment more difficult and requires more of an intervention rather than a prevention
management strategy. Both IPM in general and N management in humid regions
exemplify this situation. Managing entities with a high space-time dependence,
however, may be potentially more profitable and environmentally beneficial than
managing high spatial-low temporal-dependententities. These hypotheses remain
to be tested but appear in principle to be supported by existing studies. Perhaps
most problematic is the notion that the more complex the space-time situation, the
more difficult it may be for farmers to manage without sophisticated decision
aides. Technology transfer may be increasingly difficult as the complexity of pre-
cision agriculture increases.
One conclusion is clear: To achieve success, precision agriculture must evolve
66 FRANCIS J. PIERCE AND PETER NOWAK

from a collection of partially compatible technologies into an integrated manage-


ment system able to accommodate the diverse biophysical settings and cropping
enterprises characteristic of today’s agriculture. Precision agriculture must be
structured in ways that enable farmers to complete the basic steps of assessing vari-
ability, managing variability, and evaluating the outcome of these processes. This
enabling has not happened to the extent desired or possible and important issues
remain in this regard.
With respect to assessing variability, there is insufficient scientific guidance re-
garding exactly what properties or processes need to be measured. Although the
tools exist to measure a given entity, this does not mean that the entity needs to be
measured in assessing variability. This will be especially important as we shift the
focus to the more temporally dynamic components of crop production because it
will become as important to know when and where to look as it is to know what
to look for. For these reasons there needs to emerge a clear description of the salient
properties and processes that affect crop performance and these need to be assessed
for precision agriculture by biophysical settings and cropping systems. Lack of
these standard protocols has lead to situations in which interpretation of measured
variability is often clouded by inaccurate assessment (due to poor measurement
techniques or inadequate sampling designs) or lack of scientific understanding of
underlying causal processes. Therefore, more emphasis must be placed on both the
validity and the accuracy in variability assessment. Tools that increase assessment
accuracy of valid properties and processes while keeping costs affordable should
be a high priority for research. Sensors, both ground based and remote, should in-
crease the detail needed for precision agriculture at reduced costs, but these will
require significant improvements in software, particularly user interfaces, to fully
utilize precise data. However, sensor data also need to be integrated with advances
in software that will improve our understanding of the spatial and temporal varia-
tion in the properties and processes that regulate crop production as well as the fate
and transport of pollutants to the environment.
With regard to managing variability, the agronomic basis for most precision
management techniques is not well established due to scale incongruities. Current
management recommendations rely on traditional best management practices and
concepts such as whole-field IPM, which may or may not have site-specific cor-
relates at a finer resolution. A concerted effort should be made to design products
and management tools based on the emerging science of precision agriculture
while not forcing new enabling technologies to operate under traditional crop man-
agement strategies and guidelines. Precise management recommendations will
need to be based on measurable or predictable site-specific conditions that are con-
sistent with the capabilities of the enabling technologies. On-farm research, which
is increasing with the adoption of certain enabling technologies, offers a great op-
portunity to evaluate current and future management practices over a broad space-
time continuum not possible under traditional small-plot research.Traditional pub-
ASPECTS OF PRECISION AGRICULTURE 67

lic-sector research needs to integrate with the on-farm research network that is
rapidly emerging in the realm of precision agriculture.
With regard to evaluation, it is essential that evaluation procedures be devel-
oped that are consistent with the emerging features of precision agriculture while
not relying solely on the traditional approach used for agricultural machinery, ge-
netics, or chemicals. These evaluation procedures must involve the farmer. The
farmer is the integral part of evaluating precision agriculture because the assessed
variability must ultimately be managed on the farm. The enabling technologies,
the agronomic rules, and even the data have no value in themselves. Value is ob-
tained through the management of data on the farm. It is the data and the decisions
they lead to on the farm that are the essence of precision agriculture. Our evalua-
tion efforts need to begin with this farm management process. The impacts of pre-
cision agricultural systems will extend beyond crop production to the environment
and to the very structure of our agriculture system. Consequently, evaluation needs
to involve all sectors of agriculture. We have argued that the essence of precision
agriculture is associated with spatial and temporal variability. Evaluation also
needs to address this dimension. The critical impacts of precision agriculture will
not be found by only asking the traditional question of what is occurring; rather,
these insights will emerge when we focus on where and when it is occurring.
Precision agriculture is intuitively appealing, has captured the interest of the
agricultural sector, and offers many exciting and challenging research questions.
Lacking in this interchange, however, has been sufficient attention to underlying
scientific principles and standards. In our overview of precision agriculture, we
have attempted to point out what we believe are some unifying principles and an-
alytical deficiencies. We hope this discussion will focus future work on precision
agriculture so that both scientific and practical (agronomic, economic, and envi-
ronmental) objectives can be realized.

REFERENCES
Adams, F. (ed.) (1984). “Soil Acidity and Liming,” 2nd ed. American Society of Agronomy (ASA).
Crop Science Society of America (CSSA), and Soil Science Society of America (SSSA). Madi-
son, Wl.
Allmaras, R. R.,Wilkins, D. E., Bumside, 0. C., and Mulla, D. J. (1998). Agricultural technology and
adoption of conservation practices. I n “Advances in Soil and Water Conservation” (F. J. Pierce
and W. W. Frye, Eds.), pp. 99-158. Sleeping Bear Press, Chelsea, MI.
American Society ofAgricultural Engineers (ASAE) (1991). “Automated Agriculture for the 21st Cen-
tury. Proceedings of the 1991 Symposium, Chicago, IL.16-17 December,” ASAE Publ. No. 11-
91. ASAE, St. Joseph, MI.
Anderson, N. W., and Humburg, D. S. (1997). Application equipment for site-specificmanagement. In
“The State of Site-SpecificManagement for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.),ASA
Miscellaneous Publication, pp. 245-281. ASA, CSSA, and SSSA, Madison, WI.
Asgard, J. (1994). Soil cultivation in darkness reduced weed emergence. Acro Horr. 372, 167-177.
68 FRANCIS J. PIERCE AND PETER NOWAK

Auernhammer, H. (ed.) (1949).Global position systems in agriculture [Special issue]. Computers Elec-
tronics Agric. 11, 1-95.
Auernhammer, H., and Muhr, T.(1991).GPS in a basic rule for environment protection in agriculture.
In “Automated Agriculture for the 21st Century. Proceedings of the 1991 Symposium, Chicago,
IL, 16-17 December,” ASAE Publ. No. 11-91, pp. 395-402. ASAE, St. Joseph, MI.
Bahri, A,, Von Bargen, K., Kocher, M. F., and Bashford, L. L. (1996). Metering characteristics ac-
companying rate changes necessary for precision farming. In “Proceedings of the Third Interna-
tional Conference on Precision Agriculture. Minneapolis, MN, 23-26 June 1996 (P. C. Robert,
R. H. Rust, and W. E. Larson, Eds.),ASAMiscellaneous Publication, pp. 369-377. ASA, CSSA,
and SSSA, Madison, WI.
Bae, Y.M., Borgelt, S. C., Searcy, S. W., Schueller, J. K., and Stout, B. A. (1987). Determination of
spatially variable yield maps. ASAE Paper No. 87- 1533. ASAE, St. Joseph, MI.
Ballal, K., Krishnan, P., Kemble, J., and Issler. A. (1996).Nozzle selection and replacement based on
nozzle wear analysis. In “Proceedings of the Third International Conference on Precision Agri-
culture, Minneapolis, MN, 23-26 June, 1996” (P. C. Robert, R. H. Rust, and W. E. Larson. Eds.),
ASA Miscellaneous Publication, pp. 795-804. ASA, CSSA, and SSSA, Madison, WI.
Barnett, V.. Landau, S., Colls, J. J., Craigon, J., Mitchell, R. A. C., and Payne. R. W. (1997). Predict-
ing wheat yields: The search for valid and precise models. In “Precision Agriculture: Spatial and
Temporal Variability of Environmental Quality” (J. V. Lake, G. R. Bock, and J. A. Goode, Eds.),
pp. 79-92. Wiley, New York.
Barnhisel, R. I., Bitzer, M. J.. Grove, J. H., and Shearer, S. A. (1996). Agronomic benefits of varying
corn seed populations: A central Kentucky study. In “Proceedings of the Third International Con-
ference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust,
and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 957-965. ASA, CSSA, SSSA,
Madison, WI.
Bashford, L. L. (1993). “External Flute Seed Metering Evaluation Related to Site Specific Farming,”
ASAE Paper No. 93-8517.ASAE, St. Joseph, MI.
Bashford, L. L., Bahri, A., Von Bargen, K.,and Kocher, M. F. (1996). Variability in volume metering
devices. In “Proceedings of the Third International Conference on Precision Agriculture, Min-
neapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Mis-
cellaneous Publication, pp. 693-702. ASA, CSSA, and SSSA, Madison, WI.
Baumgardener, M. F., Kristof, S., Johannsen, C. J., and Zachary, A. (1970). Effects of organic matter
on the multispectral properties of soils. Indiana Acad. Sci. 79,413-422.
Bausch, W. C., Duke, H. R., and Iremonger, J. C. (1996).Assessment of plant nitrogen in irrigated corn.
In “Proceedings of the Third International Conference on Precision Agriculture, Minneapolis,
MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.). ASA Miscellaneous
Publication, pp. 23-32. ASA, CSSA, and SSSA, Madison, WI.
Beckett, P. H. T.. and Webster, R. (1971). Soil variability: A review. Soils Fertilizers 34, 1-15.
Bell, J. C., Cunningham, R. L., and Havens, M. W. (1992). Calibration and validation of a soil-land-
scape model for predicting soil drainage class. Soil Sci. SOC.Am. J. 56, 1860-1866.
Bell, J. C., Cunningham, R. L., and Havens, M. W. (1994).Soil drainage class probability mapping us-
ing a soil-landscape model. Soil Sci. SOC.A n J. 58,464-470.
Bell, J. C., Butler, C. A., and Thompson, J. A. (1995). Soil-terrain modeling for site-specific manage-
ment. In “Proceedings of the Second International Conference on Site Specific Management for
Agricultural Systems, Bloomington/Minneapolis,MN, 27-30 March 1994” (P. C. Robert, R. H.
Rust, and W. E. Larson. Eds.), ASA Miscellaneous Publication, pp. 209-227. ASA, CSSA, and
SSSA, Madison, WI.
Benedict, H. M., and Swidler, R. (1961).Nondestructive method for estimating chlorophyll content of
leaves. Science 133,2015-2016.
Berry, J. K. (1993).“Beyond Mapping: Concepts, Algorithms, and Issues in GIS.” GIS World Books,
Ft. Collins, CO.
ASPECTS OF PRECISION AGRICULTURE 69

B e y , J. K. (1995). “Spatial Reasoning for Effective GIS.” CIS World Books, Ft. Collins, CO.
BIOS (1997). “Precision Agriculture 1997.” Bios Sci., Oxford, UK.
Birrell, S. J., Sudduth, K. A., and Kitchen. N. R. (1996). Nutrient mapping implications of short-range
variability. In “Proceedings of the Third International Conference on Precision Agriculture, Min-
neapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Mis-
cellaneous Publication, pp. 206-216. ASA, CSSA, and SSSA, Madison, WI.
Black, C. A. (1993). “Soil Fertility Evaluation and Control.” Lewis, Boca Raton, FL.
Blackmer, A. M., and White, S. E. (1996). Remote sensing to identify spatial patterns in optimal rates
of nitrogen fertilization. In “Proceedings of the Third International Conference on Precision Agri-
culture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 33-41. ASA, CSSA, and SSSA, Madison, WI.
Blevins, R. L., Lal, R., Doran. J. W., Langdale, G. W., and Frye, W. W. (1998). Conservation tillage for
erosion control and soil quality. In “Advances in Soil and Water Conservation” (F. J. Pierce and
W. W. Frye, Eds.), pp. 5 1-68. Sleeping Bear Press, Chelsea, MI.
Bock, B. R., Kelley, K. R., and Meisinger, J. J. (1992). Predicting N fertilizer needs for corn in humid
regions: Summary and future directions. In “Predicting N Fertilizer Needs for Corn in Humid Re-
gions” (B. R. Bock and K. R. Kelly, Eds.), Bull. No. Y-226, TVAINFERC-92I2, pp. 115-127. Na-
tional Fertilizer and Environmental Research Center, Tennessee Valley Authority, Muscle Shoals,
AL.
Boehlje, M. (1994). Information: What is the public role? Staff Paper No. 94-17. Department of Agri-
cultural Economics, Purdue University. West Lafayette, IN.
Borgelt, S. C., Searcy. S. W., Stout, B. A., and Mulla, D. J. (1994). Spatially variable liming rates: A
method for determination. Trans. Am. Soc. Agric. Eng. 37, 1499-1057.
Bouma. J. (1997). Precision agriculture: Introduction to the spatial and temporal variability of envi-
ronmental quality. In “Precision Agriculture: Spatial and Temporal Variability of Environmental
Quality” (J. V. Lake, G. R. Bock, and J. A. Goode, Eds.), pp. 5-13. Wiley. New York.
Brown, R. B.. and Steckler, J.-R G. A. (1995). Prescription maps for spatially variable herbicide ap-
plication in no-till corn. Trans. Am. SOC.Agric. Eng. 38, 1659-1666.
Brown, R. B., Andersen. G . W., Proud, B., and Steckler. J. P.(1990). Herbicide application control us-
ing GIS weed maps, ASAE Paper No. 90-1061. ASAE, St. Joseph, MI.
Brubaker, S. C., Jones, A. J., Lewis, D. T., and Frank, K. (1993). Soil properties associated with land-
scape position. Soil Sci. Soc. Am. J. 57,235-239.
Bullock, D. G., Bullock, D. S., Nafziger, E. D., Doerge, T., Paszkiewicz, S., Carter, P., and Peterson,
T. A. (1998). Does variable rate seeding of corn pay? Agron. J., 90,830-836.
Bundy, L. G., Schmitt, M. A., and Randall, G. W. (1992). Predicting N fertilizer needs for corn in hu-
mid regions: Advances in the upper Midwest. In “Predicting N Fertilizer Needs for Corn in Hu-
mid Regions” (B. R. Bock and K. R. Kelly, Eds.). Bull. No. Y-226, TVAINFERC-92I2, pp. 73-
89. National Fertilizer and Environmental Research Center, Tennessee Valley Authority, Muscle
Shoals, AL.
Burrough, P. A. (1986). “Principles of Geographic Information Systems for Land Resource Assess-
ment.” Clarendon, Oxford.
Cahn, M. D., Hummel, J. W., and Brouer, B. H. (1994). Spatial analysis of soil fertility for site-specif-
ic crop management. Soil Sci. SOC.Am. J. 58, 1240-1248.
Camp, C. R., and Sadler, E. J. (1994). Center pivot irrigation system for site-specific water and nutri-
ent management, ASAE Paper No. 94-1586. ASAE, St. Joseph, MI.
Camp, C. R., Sadler, E. J., Evans. D. E., Usery, L. J., and Omary, M. (1996). Center pivot irrigation
system for site-specific water and nutrient management, ASAE Paper No. 96-2077. ASAE, St.
Joseph, MI.
Cam, P. M., Carlson, G. R., Jacobsen, J. S., Nielsen, G . A., and Skogley, E. 0. (1991). Farming soils,
not fields: A strategy for increasing fertilizer profitability. J. Prod. Agric. 4,57-61.
Cattanach, A., Franzen, D., and Smith, L. (1996). Grid soil testing and variable rate fertilizer applica-
70 FRANCIS J. PIERCE AND PETER NOWAK

tion effects on sugarbeet yield and quality. I n “Proceedings of the Third International Conference
on Precision Agriculture, Minneapolis, MN. 23-26 June 1996” (P. C. Robert, R. H. Rust, and
W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 1033-1038. ASA, CSSA, and SSSA,
Madison, WI.
Chaplin, J., Roytburg, E., and Kaplan, J. (1995). Measuring the spatial performance of chemical ap-
plicators. In “Proceedings of the Second International Conference on Site Specific Management
for Agricultural Systems. BloomingtonlMinneapolis. MN, 27-30 March 1994” (P. C. Robert,
R. H. Rust, and W. E. Larson, Eds.),ASAMiscellaneousPublication, pp. 651-669. ASA, CSSA,
and SSSA, Madison, WI.
Christensen, D. R., Bricker, C. E., and Zehr,G. L. (1998). Effect of lime applied to high pH soils on
yield of sugarbeet, corn. soybean, and navy bean. In ‘The 1997 Research Report of the Saginaw
Valley Bean &Beet Research Farm and Related Bean-Beet Research,” pp. 11-19. Michigan State
Univ. Agricultural Experiment Station, East Lansing, MI.
Ciha, A. J. (1984). Slope position and grain yield of soft white winter wheat. Agron. J. 76, 193-196.
Clark, R. L. (1996).A comparison of rapid GPS techniques for topographic mapping. In “Proceedings
of the Third International Conference on Precision Agriculture, Minneapolis, MN, 23-26 June
1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp.
651-662. ASA, CSSA, and SSSA. Madison, WI.
Clark, R. L., and McGucken,R. L. (1996). Variable rate application technology: An overview.In “Pro-
ceedings of the Third International Conference on Precision Agriculture, Minneapolis, MN, 23-
26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publica-
tion, pp. 855-862. ASA, CSSA, and SSSA, Madison, WI.
Colliver, C. T., Maxwell, B. D., Tyler, D. A., Roberts, D. W., and Long, D. S. (1996). Georeferencing
wild oat infestations in small grains: Accuracy and efficiency of three weed survey techniques. In
“Proceedings of the Third International Conference on Precision Agriculture, Minneapolis, MN,
23-26 June 1996” (P. C. Robert, R. H.Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publi-
cation, pp 453-463. ASA, CSSA, and SSSA, Madison, WI.
Cox, F. R. (1994). Current phosphorus availability indices: Characteristics and shortcomings. In “Soil
Testing: Prospects for Improving Nutrient Recommendations” (J. L. Havlin and J. S. Jacobsen,
Eds.), SSSA Special Publ. No. 40,pp. 101-113. SSSA, Madison, WI.
Cox, W. J. (1996). Whole-plant physiological and yield responses of maize to plant density. Agron. J.
88,489-496.
Cressie, N. A. C. (1991). “Statistics for Spatial Data.” Wiley, New York.
Crookston, R. K. (1996). Using decision cases to enhance technology transfer in agriculture. In “Pro-
ceedings of the Third International Conference on Precision Agriculture, Minneapolis, MN, 23-
26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publica-
tion, pp. I 117-1 122.ASA, CSSA, and SSSA, Madison, WI.
Dahnke,W. C., and Olson, R. A. (1990).Soil test correlation, calibration,and recommendation.In “Soil
Testing and Plant Analysis: Third Edition” (R. L. Westerman, Ed.), SSSA Book Series No. 3. pp.
45-71. SSSA, Madison, WI.
Daniels, R. B., Gilliam, J. W., Cassel, D. K., and Nelson, L. A. (1985). Soil erosion class and landscape
position in North Carolina piedmont. Soil Sci. SOC.Am. J. 49,99 1-995.
De Baerdemaeker, J., Delcroix, R., and Lindemans, P. (1985). Monitoring the grain flow in combines.
In “Agri-Mation I , Proceedings of the Agri-Mation I Conference and Exposition, Chicago, JL,
Feb. 25-28, 1985,” pp. 329-338. ASAE Publ. 01-85, ASAE, St. Joseph, MI.
Denton, B. (1996). Precision information: How can it pay? In “Proceedings of the 1996 Information
Agriculture Conference, Urbana, IL, July 30-August I , 1996,” pp. 41ff. Potash and Phosphate
Institute and the Foundation for Agronomic Research, Norcross, GA.
Dessaint, R.. Chadoeuf, R., and Barralis, G . (1991). Spatial pattern analysis of weed seeds in cultivat-
ed soil seedbank. J. Appl. Ecol. 28,721-730.
Devlin. D. L., Fjell, D. L., S h y e r , J. P., Gordon, W. B., Marsh, B. H., Maddux, L. D., Martin, V. L.,
ASPECTS OF PRECISION AGRICULTURE 71

and Duncan, S. R. (1995). Row spacing and seeding rates for soybean in low and high yielding
environments. J. Prod. Agric. 8,215-222.
Divis, D. E. (1998, February). OMB attacks national DGPS. GPS World, 14-16.
Duff, P. (1993). Detectspray system. Pmc. Weed Sci. SOC.Am. 33,45.
Duvick, D. N. (1996). Plant breeding, an evolutionary concept. Crop Sci. 36,539-548.
Elliot, C. (1987). “Fertilizing-Blending and Spreading On-the-Go Using Computerized Maps and
Radar Guidance,” Society of Automotive Engineers Paper No. 87-1676. Society of Automotive
Engineers, Warrendale,PA.
Ellsbury, M. M., Woodson, W. D., Clay, S. A., and Carlson, C. G. (1996). Spatial characterization of
corn rootworm populations in continuous and rotated corn. In “Proceedings of the Third Interna-
tional Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert,
R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 487-494. ASA, CSSA,
and SSSA, Madison, WI.
Engel. B. A., and Gaultney, L. D. (1990). “Environmentally Sound Agricultural Production Systems
Through Site-Specific Farming,” ASAE Paper No. 90-2566. ASAE, St. Joseph, MI.
Environmental Systems Research Institute (ESRI) (1997). “Understanding GIs: The ARClINFO
Method.” ESRIlWiley, Redlands, CA/New York.
Ess, D. R., Joern, B. C., and Hawkins, S. E. (1996). Development of a precision application system for
liquid animal manure. In “Proceedings of the Third International Conference on Precision Agri-
culture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W.E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 863-870. ASA, CSSA, and SSSA, Madison, WI.
Evans, R. G., Han, S., Kroeger, M. W., and Schneider, S. M. (1996). Precision center pivot irrigation
for efficient use of water and nitrogen. In “Proceedings of the Third International Conference on
Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E.
Larson. Eds.),ASA MiscellaneousPublication, pp. 75-84. ASA, CSSA, and SSSA, Madison, WI.
Everett, M. W.. and Pierce, F. J. (1996). Variability of corn yield and soil profile nitrates in relation to
site-specific N management. In “Proceedings of the Third International Conference on Precision
Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson,
Eds.), ASA Miscellaneous Publication, pp. 45-53. ASA, CSSA, an SSSA, Madison, WI.
Fairchild, D. (1993). Working group report on the profitability of soil specific crop management. In
“Proceedings of Soil SpecificCrop Management:A Workshop on Research and Development Is-
sues, 14-16April 1992, Minneapolis,MN” (P. C. Robert, R. H.Rust and W. E. Larson, Eds.), pp.
245-253. ASA, CSSA, and SSSA, Madison, WI.
Farm Industry News (1994). Win-win for the environment. Farm Industry News 27(12), 34-35.
Fehr, W. R. (Ed.) (1987). “Principles of Cultivar Development, Volume 1: Theory and Technique.”
Macmillan, New York.
Felton, W. L., and McCloy, K. R. (1992). Spot spraying. Agric. Eng. 73,9-12.
Felton, W. L., Doss, A. F., Nash, P. G., and McCloy, K. R. (1991). A microprocessor controlled tech-
nology to selectively spot spray weeds. In “Automated Agriculture for the 21st Century. Pro-
ceedings of the 1991 Symposium, Chicago, IL, 16-17 December,” ASAE Publ. NO. 11-91, pp.
427-432. ASAE, St. Joseph, MI.
Ferguson, R. B., Gotway, C. A., Hergert, G. W.. and Peterson, T. A. (1996). Soil sampling for site-spe-
cific nitrogen management. In “Proceedings of the Third International Conference on Precision
Agriculture, Minneapolis, MN, 23-26 June 1996“ (P. C. Robert. R. H. Rust, and W. E. Larson,
Eds.), ASAMiscellaneousPublication, pp. 13-22. ASA, CSSA, and SSSA, Madison, WI.
Fiez, T. E., and Miller, B. D. (1995). Varying winter wheat seeding rates among landscape positions.
J. Prod. Agric. 8,346-350.
Fiez, T. E., Miller, B. C., and Pan, W. L. (1994a). Assessment of spatially variable nitrogen fertilizer
management in winter wheat. J. Prod. Agric. 7, 17-18.86-93.
Fiez, T.E., Miller, B. C., and Pan, W. L. (1994b). Winter wheat yield and grain protein across varied
landscape positions. Agron. J. 86, 1026-1032.
72 FRANCIS J. PIERCE AND PETER NOWAK

Fiez, T. E., Miller, B. C., and Pan, W. L. (1995).Nitrogen efficiency analysis of winter wheat among
landscape positions. Soil Sci. SOC.Am. J. 59,1666-1671.
Fixen, P. E. (1998).Research needs for site-specific nutrient management to benefit agriculture. Berm
Cr0~~82(1),
16-18.
Fleischer, S. J., Weisz, R., Smilowitz, Z., and Midgarden, D. (1997).Spatial variation in insect popu-
lations and site-specific integrated pest management. In ‘The State of Site-Specific Management
for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA Miscellaneous Publication, pp. 101-
130.ASA, CSSA, and SSSA, Madison, WI.
Forcella, F.(1993).Value of managing within-field variability. In “Proceedings of Soil Specific Crop
Management: A Workshop on Research and Development Issues, 14-16 April 1992,Minneapo-
lis, MN” (P.C. Robert, R. H. Rust, and W. E. Larson, Eds.), pp. 125-132.ASA, CSSA, and SSSA,
Madison, WI.
Fortin, M.-C.. and Pierce, F. J. (1998).Toward an agricultural information system to maximize value
in agricultural data. In “Privatization of Information and Agricultural Industrialization” (S. A.
Wolf, Ed.), pp. 95-104.CRC Press, Boca Raton, FL.
Franzen, D. W., and Peck, T. R. (1995).Spatial variability of plant analysis calcium and magnesium
levels before and after liming. Commun. Soil Sci. Plan; Anal. 26,2263-2277.
Franzen, D.W.. Cihacek, L. J., and Hofrnan. V. L. (1996).Variability of soil nitrate and phosphate un-
der different landscapes. In “Proceedings of the Third International Conference on Precision Agri-
culture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 52 1-529.ASA, CSSA, and SSSA, Madison, WI.
Frazier, B. E., Walters, C. S., and Perry, E. M. (1997).Role of remote sensing in site-specific man-
agement. In “The State of Site-Specific Management for Agriculture” (F. J. Pierce and E. J.
Sadler, Eds.), ASA Miscellaneous Publication, pp. 149-160.ASA, CSSA, and SSSA, Madi-
son, WI.
Frye. W. W., Blevins, R. L., Smith, M. S., Corak. S. J., and Varco, J. J. (1988).Role of annual legume
cover crops in efficient use of water and nitrogen. In “Cropping Strategies for Efficient Use of Wa-
ter and Nitrogen” (W. L. Hargrove, Ed.), ASA Special Publ. No. 51,pp. 129-152.ASA, CSSA,
and SSSA, Madison, WI.
Gerhards, R., Wyse-Pester, D. Y., and Mortensen, D. A. (1996).Spatial stability of weed patches in
agricultural fields. In “Proceedings of the Third International Conference on Precision Agricul-
ture, Minneapolis, MN, 23-26 June 1996 (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 495-504.ASA, CSSA, and SSSA, Madison, WI.
Giles, D. K.,Henderson, G. W., and Funk, K. (1996).Digital control of flow rate and spray droplet size
from agricultural nozzles for precision chemical application. In “Proceedings of the Third Inter-
national Conference on Precision Agriculture, Minneapolis. MN, 23-26 June 1996” (P. C. Robert,
R. H.Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 729-738.ASA, CSSA,
and SSSA, Madison, WI.
Goedeken, M., Johnson, G.. and Raun, B. (1998).Expectations of precision phosphate management.
Berrer Crops 82(I). 28-31.
Goodchild, M. F., Parks, B. 0..and Steyaert, L. T. (Eds.) (1993).“Environmental Modeling with GIS.”
Oxford Univ. Press, New York.
Goovaerts, P. (1997).“Geostatistics for Natural Resource Evaluation.’’Oxford Univ. Press, New York.
Goovaerts, P. (1999).Geostatistics in soil science: State-of-the-art and perspectives. Geodenna, 89
(l-2),1-45.
Gotway, C. A., Ferguson, R. B., and Hergert, G. W. (1996a).The effects of mapping and scale on vari-
able-rate fertilizer recommendations for corn. In “Proceedings of the Third International Confer-
ence on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust,
and W. E. Larson, Eds.), ASAMiscellaneous Publication, pp. 321-330.ASA, CSSA, and SSSA.
Madison, WI.
ASPECTS OF PRECISION AGRICULTURE 73

Gotway. C. A., Ferguson, R. B., Hergert, G. W., and Peterson, T.A. (1996b). Comparison of kriging
and inverse-distancemethods for mapping soil parameters. Soil Sci. SOC.Am. J. 60,1237-1247.
Groffman, P. M. (1997). Ecological constraints on the ability of precision agriculture to improve the
environmental performance of agriculturalproduction systems. In “PrecisionAgriculture: Spatial
and Temporal Variability of Environmental Quality” (J. V. Lake, G. R. Bock, and J. A. Goode,
Eds.), pp. 52-64. Wiley, New York.
Guyer, D. E., Miles, G. E., Schreiber, M. M.,Mitchell, 0. R., and Vanderbilt, V. 0. (1986). Machine
vision and image processing for plant identification. Trans.Am. SOC.Agric. Eng. 29,1500-1507.
Haggar. R. J., Stent, C. J., and Issac, S. (1983). Aprototype hand-held patch sprayer for killing weeds,
activated by spectral differences in crop weed canopies. J. Agric. Eng. Res. 28,349-358.
Halvorson, G . A., and Doll. E. C. (1991). Topographic effects on spring wheat yields and water use.
Soil Sci. SOC.A m J. 55, 1680-1685.
Han, S., Schneider, S. M., Evans, R. G., and Rawlins, S. L. (1996). Spatial variability of soil proper-
ties on two center pivot irrigated fields. In “Proceedingsof the Third International Conference on
Precision Agriculture, Minneapolis, MN, 23-26 June 1 9 9 6 (P. C. Robert, R. H. Rust, and W. E.
Larson, Eds.), ASA Miscellaneous Publication, pp. 97-106. ASA, CSSA, and SSSA, Madison,
WI.
Hanna, A. Y.,Harlan, P. W., and Lewis, D. T. (1982). Soil available water as influenced by landscape
position and aspect. Agron. J. 74,999-1004.
Hanson, L. D., Robert, P.C.. and Bauer, M. (1995). Mapping wild oat infestations using digital im-
agery for site-specific management. In “Proceedings of the Second International Conference on
Site Specific Management for Agricultural Systems, BloomingtonlMinneapolis, MN, 27-30
March 1994” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication,
pp. 495-503. ASA, CSSA, and SSSA, Madison, WI.
Hartmann, K. M., and Nezadal, W. (1990). Photocontrol of weeds without herbicides. Nurumis-
senschujien 77,158-163.
Hatfield, J. L., and Pinter, P. J., Jr. (1993). Remote sensing for crop protection. Crop Protec:ion 12,
403-41 3.
Hatfield, J. L., and Stewart, B. A. (Eds.) (1997). “Animal Waste Utilization: Effective Use of Manure
as a Soil Resource.” Ann Arbor Press, Chelsea, MI.
Heisel, T., Andreasen. C., and Ersboll, A. K. (1996). Annual weed densities can be mapped with krig-
ing. Weed Res. 36,325-337.
Henry, J. L.. and MacDonald,K. B. (1978).The effects of soil and fertilizernitrogen and moisture stress
on yield, oil and protein content of rape. Can. J. Soil Sci. 58,303-310.
Hergert. G. W., Ferguson. R. B., and Shapiro, C. A. (1995a). Fertilizer suggestions for corn, Universi-
ty of Nebraska NebGuide G74-174 (Revised). Univ. of Nebraska, Lincoln.
Hergert. G. W., Ferguson, R. B., Shapiro, C. A., Penas, E. J., and Anderson, F. B. (1995b). Classical
statistical and geostatistical analysis of soil nitrate-N spatial variability. In “Proceedings of the
Second InternationalConference on Site Specific Management for Agricultural Systems. Bloom-
ington/Minneapolis,MN. 27-30March 1994” (P.C. Robert, R. H.Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 175-186. ASA, CSSA, and SSSA, Madison, WI.
Hergert, G. W., Ferguson, R. B., Gotway, C. A., and Peterson, T.A. (1996). The impact of variable rate
N application on N use efficiency of furrow irrigated corn. In “Proceedingsof the Third Interna-
tional Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert,
R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 389-397. ASA, CSSA,
and SSSA, Madison, WI.
Hergert, G. W., Pan, W. L., Huggins. D. R., Grove, J. H., and Peck, T. R. (1997). Adequacy of current
fertilizer recommendations for site-specific management. In “The State of Site-SpecificManage-
ment for Agriculture” (F. J. Pierce and E. J. Sadler. Eds.), ASA Miscellaneous Publication, pp.
283-300. ASA, CSSA, and SSSA, Madison, WI.
74 FRANCIS J. PIERCE AND PETER NOWAK

Hesterman, 0. B. (1988). Exploiting forage legumes for nitrogen contributions in cropping systems.
In “Cropping Strategies for Efficient Use of Water and Nitrogen” (W. L. Hargrove, Ed.), ASA Spe-
cial Publ. No. 51, pp. 155-166. ASA, CSSA, and SSSA, Madison, WI.
Hicks, D. R., Lueschen, W. E.. and Ford, J. H. (1990). Effect of stand density and thinning on soybean.
J. Prod. Agric. 3,587-590.
Hillel, D. ( 1 990). Role of irrigation in agricultural systems. In ‘‘Irrigation of Agricultural Crops”
(B. A. Stewart and D. R. Nielsen, Eds.), Agronomy Monograph No. 30, pp. 5-30. ASA, CSSA,
and SSSA, Madison, WI.
Hills, F. J., and Ulrich, A. (1971). Nitrogen nutrition. In “Advances in Sugarket Production: Princi-
ples and Practices” (R. T.Johnson, J. T. Alexander, G.E. Rush, and G. R. Hawkins, Eds.), pp.
11 1-135. Iowa State Univ. Press, Ames.
Hoeft, R. G., Brown, H. M.. Mengel, D., Eckert, D. J., and Vitosh, M. L. (1992). Predicting N fertiliz-
er needs for corn in humid regions: Advances in the Midwest. In “Predicting N Fertilizer Needs
for Corn in Humid Regions” (B. R. Bock and K. R. Kelly. Eds.), Bull. No. Y-226. TVAINFERC-
92/2, pp. 90-104. National Fertilizer and Environmental Research Center, Tennessee Valley Au-
thority, Muscle Shoals, AL.
Hoffmann-Wellenhof, B.,Lightenegger, H., and Collins, J. (1994). “GPS: Theory and Practice,” 3rd
ed. Springer-Verlag. New York.
Hollands, K. R. (1996). Relationship of nitrogen and topography. In “Proceedings of the Third Inter-
national Conference on Precision Agriculture, Minneapolis, MN. 23-26 June 1996” (P. C. Robert,
R. H. Rust, and W. E. Larson, Eds.), ASAMiscellaneous Publication, pp. 3-12. ASA, CSSA, and
SSSA, Madison, WI.
Holmberg, M. (1998). Paths to precision. Successful Farming %(5), 33-34.
Holmes. W. (1993). Prescription farming. In “ P r o c d m g s of Soil Specific Crop Management: A Work-
shop on Research and Development Issues, 14-16 April 1992, Minneapolis, MN” (P. C. Robert,
R. H. Rust, and W. E. Larson, Eds.). pp. 311-316.ASA, CSSA, andSSSA, Madison, WI.
Holt, D. A. (1985). Computers in production agriculture. Science 228,422-427.
Holt, D. A., and Sonka, S. T. (1995). Virtual agriculture: Developing and transferring agricultural tech-
nology in the 21st century. In “Proceedings of the Second International Conference on Precision
Agriculture, Minneapolis, MN, 27-30 March 1994” (P. C. Robert, R. H. Rust, and W. E. Larson,
Eds.), ASA Miscellaneous Publication, pp. 883-898. ASA, CSSA, and SSSA, Madison, WI.
Hornbaker, R. H. (1996). Economic assessment of precision farminglsite specific management sys-
tems: Current research results, issues and future directions. In “Proceedings: 1996 Information
Agriculture Conference, University of Illinois, Urbana. IL, July 30-August 1.” pp. 78-79. Potash
and Phosphate Institute, Brookings, SD.
Howell, T. (1990). Relationships between crop production and transpiration, evapotranspiration, and
irrigation. In “Irrigation of Agricultural Crops” (B. A. Stewart and D. R. Nielsen. Eds.), Agrono-
my Monograph No. 30, pp. 391-434. ASA, CSSA, and SSSA, Madison, WI.
Isaaks, E. H., and Srivastava, R. M. (1989). “Applied Geostatistics.” Oxford Univ. Press, New York.
Johnson, G. A., Mortensen, D. A., Young, L. J., and Martin, A. R.(1995). The stability of weed seedling
population models and parameters in eastern Nebraska corn (&a mays L.) and soybean (Glycine
max L.) fields. Weed Sci. 43,604-61 1.
Johnson, G. A., Cardina, J., and Mortenson. D. A. (1997). Site-specific weed management: Current and
future directions. In ‘“The State of Site-Specific Management for Agriculture” (F. J. Pierce and
E. J. Sadler. Eds.). ASA Miscellaneous Publication, pp. 101-130. ASA, CSSA, and SSSA, Madi-
son, WI.
Jones, A. J., Meilke, L. N., Bartles, C. A., and Miller, C. A. (1989). Relationship of landscape position
and properties to crop production. J. Soil Water Conservation 44,328-332.
Kachanoski, R. G.,Fairchild, G. L., and Beauchamp, E. G.(1996). Yield indices for corn response to
applied fertilizer: Application in site-specific crop management. In “Proceedings of the Third In-
ASPECTS OF PRECISION AGRICULTURE 75

ternational Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C.
Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 425-432. ASA,
CSSA, and SSSA, Madison, WI.
Kaplan. E. D. (Ed.) (1996). “Understanding GPS: Principles and Applications.” Artech House, Boston.
Kaplan, J., and Chaplin, J. (1996). Estimation of quality of fertilizer distribution. In “Proceedings of
the Third International Conference on Precision Agriculture, Minneapolis, MN, 23-26 June
1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp.
871 -883. ASA, CSSA, and SSSA, Madison, WI.
Kellogg, C. E. (1961). “Soil Interpretation in the Soil Survey,” in-service publication. United States
Department of Agriculture, Soil Conservation Service, Washington, DC.
Kennedy, M. (1996). “The Global Position System and GIS.” Ann Arbor Press, Chelsea, MI.
Khakural, B. R., Robert, P. C., and Koskinen, W. C. (1995).Runoff and leaching of alachlor under con-
ventional and soil-specific management. Soil Use Management 10, 158-164.
Khakural, B. R.. Robert, P.C., and Mulla, D. J. (1996). Relating cornlsoybean yield to variability in
soil and landscape characteristics. In “Proceedings of the Third International Conference on Pre-
cision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Lar-
son, Eds.), ASA Miscellaneous Publication, pp. 117-128. ASA, CSSA, and SSSA, Madison, WI.
King, B. A., Brady, R. A., McCann, I. R., and Stark, J. C. (1995).Variable rate water application through
sprinkler irrigation. In “Proceedings of the Second International Conference on Site Specific Man-
agement for Agricultural Systems, BloomingtonlMinneapolis,MN, 27-30 March 1994” (P. C.
Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 485-493. ASA,
CSSA, and SSSA, Madison, WI.
King, B. A,. Stark, J. C., McCann, I. R., and Westermann, D. T. (1996). Spatially varied nitrogen ap-
plication through a center pivot irrigation system. In “Proceedings of the Third International Con-
ference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust,
and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 85-94. ASA, CSSA, and SSSA,
Madison, WI.
Kirk, I. W.. and Tom, H. H. (1996). Precision GPS flow control for aerial spray applications. In “Pro-
ceedings of the Third International Conference on Precision Agriculture, Minneapolis, MN, 23-
26 June 1996” (P. C. Robert, R. H.Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publica-
tion, pp. 815-817. ASA, CSSA, and SSSA, Madison, WI.
Kitchen, N. R., Hughes, D. F., Sudduth, K. A., and Birrell, S. J. (1995). Comparison of variable rate to
single rate nitrogen fertilizer application: Corn production and residual soil NO,-N. In “Pro-
ceedings of the Second International Conference on Site Specific Management for Agricultural
Systems, BloomingtonlMinneapolis, MN, 27-30 March 1994” (P. C. Robert, R. H. Rust, and
W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 427-441. ASA. CSSA, and SSSA,
Madison, WI.
Kohls, C. (1996, March). Costs of precision: Here’s a sampling of equipment costs to aid in your pre-
cision planning. Farm Chem. 159,21-26.
Kreznor, W. R., Olson, K. R., Banwart, W. L., and Johnson, D. L. (1989). Soil, landscape, and erosion
relationships in a northwest Illinois watershed. Soil Sci. Soc. Am. J. 53, 1763-1771.
Kropff, M. J., Wallinga, J., and Lotz, L. A. P. (1997). Modeling for precision weed management. In
“Precision Agriculture: Spatial and Temporal Variability of Environmental Quality” (J. V. Lake,
G. R. Bock, and J. A. Goode, Eds.), pp. 182-200. Wiley, New York.
Lake, J. V., Bock, G.R., and Goode, J. A. (Eds.) (1997). “Precision Agriculture: Spatial and Temporal
Variability of Environmental Quality.” Wiley, New York.
Langdale, G.W., Box, J. E., Jr., Leonard, R. A., Barnett, A. P., and Fleming, W. G.(1979).Corn yield
reduction on eroded southern piedmont soils. J. Soil WaferConservation 34,226-228.
Lange, A. F. (1996). Centimeter accuracy differential GPS for precision agriculture applications. I n
“Proceedings of the Third International Conference on Precision Agriculture, Minneapolis, MN,
76 FRANCIS J. PIERCE AND PETER NOWAK

23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publi-
cation, pp. 675-680. ASA, CSSA, and SSSA, Madison, WI.
Larson, W. E., and Robert, P. C. (1991). Fanning by soil. In “Soil Management for Sustainability” (R.
La1 and F. J. Pierce, Eds.), pp. 103-1 12. Soil and Water Conservation Society, Ankeny, IA.
Larson, W. E., Pierce, F. J.. and Dowdy, R. H. (1983). The threat of soil erosion to long-term crop pro-
duction. Science 219,458-465.
Larson, W. E., Lamb, J. A., Khakural, B. R., Fergeson, R. B., and Rehm, G.W. (1997). Potential of
site-specific management for nonpoint environmental protection. In “The State of Site-Specific
Management for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA Miscellaneous Publica-
tion, pp. 337-367. ASA, CSSA, and SSSA, Madison, WI.
Larson, W. E., Mausbach, M. J., Schmidt, B. L., and Crosson, P. (1998). Policy and government pro-
grams in soil and water conservation. In “Advances in Soil and Water Conservation” (F. J. Pierce
and W. W. Frye, Eds.), pp. 195-218. Sleeping Bear Press, Chelsea, MI.
Laslett, G.M.. and McBratney, A. B. (1990). Further comparison of spatial methods for predicting soil
pH. SoilSci. SOC.Am. J. 54,1553-1558.
Laslett, G.M., McBratney, A. B., Pahl, P. J., and Hutchinson, M. F. (1987). Comparison of several spa-
tial prediction methods for soil pH. J. Soil Sci. 38,325-341.
Lass, L. W., and Callihan, R. H.(1993). GPS and GIS for weed surveys and management. Weed Sci.
I , 249-254.
Leick, A. (1995). “GPS Satellite Surveying,’’ 2nd ed. Wiley, New York.
Lenz, D. (1996). Calculating profitability of grid soil sampling and variable rate fertilizing for sugar
beets. In “Proceedings of the Third International Conference on Precision Agriculture, Min-
neapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASAMis-
cellaneous Publication, pp. 945-947. ASA, CSSA, and SSSA, Madison, WI.
Lilleboe.. D. (1996, February). Will it pay? Sugarbeet Grower; 18-20.
Lindstrom, M. J., and Voorhees, W. B. (1994). Responses of temperate crops in North America to soil
compaction. In “Soil Compaction in Crop Production” (B. D. Soane and C. Van Ouwerkerk, Eds.),
pp. 265-286. Elsevier. New York.
Linsley, C. M., and Bauer, F. C. (1929). Test your soil for acidity, Circular No. 346. University of Illi-
nois, College of Agriculture and Agricultural Experiment Station, Urbana.
Long, D. S., Carlson, G.R., and &Gloria, S . D. (1995). Quality of field management maps. In “Pro-
ceedings of the Second International Conference on Site Specific Management for Agricultural
Systems, BloomingtonlMinneapolis, MN, 27-30 March 1994” (P. C. Robert, R. H. Rust, and
W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 251-271. ASA, CSSA, and SSSA,
Madison, WI.
Long, D. S., Carlson, G.R.,and Nielsen, G.A. (1996). Cost analysis of variable rate application of ni-
trogen and phosphorus for wheat production in northern Montana. In “Proceedings of the Third
International Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C.
Robert, R. H. Rust, and W. E. Larson, Eds.),ASA Miscellaneous Publication, pp. 1019-1031.
ASA, CSSA, and SSSA, Madison, WI.
Lowenberg-DeBoer, J. (1997, November). Bumpy road to adoption of precision agriculture. Purdue
Agricultural Economics Report. Purdue University Cooperative Extension Service, West Lafay-
ette, IN.
Lowenberg-&Boer, J., and Boehlje, M. (1996). Revolution, evolution or dead-end: Economic per-
spectives on precision agriculture. In “Proceedings of the Third International Conference on Pre-
cision Agriculture, Minneapolis, MN. 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Lar-
son, Eds.), ASA Miscellaneous Publication, pp. 923-944. ASA, CSSA, and SSSA, Madison, WI.
Lowenberg-DeBoer.J., and Swinton, S. M. (1997). Economics of site-specific management in agronomic
crops. In ‘The State of Site-SpecificManagement for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.),
ASA Miscellaneous Publication, pp. 369-396. ASA, CSSA, and SSSA, Madison, WI.
ASPECTS OF PRECISION AGRICULTURE 77
Luellen, W. R. (1985). Fine-tuned fertility: Tomorrow’s technology here today. Cmps Soils 38(2), 18-
22.
Magdoff. F. R., Jokela, W. E., Fox, R. H., and Griffin, G . F. (1990). A soil test for nitrogen availabili-
ty in the northeastern United States. Commun. Soil Sci. Planr Anal. 21, 1103-11 15.
Mallarino. A. P., Hinz, P. N., and Oyarzabla, E. S. (1996). Multivariate analysis as a tool for interpret-
ing relationships between site variables and crop yield. In “Proceedings of the Third Internation-
al Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (p. C. Robert,
R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 15 1-158. ASA, CSSA,
and SSSA, Madison, WI.
Malzer, G. L., Vetsch, J. A., Robert, P. C., and Huggins, D. R. (1995). Nitrogen specific management
by soil condition: Effects of nitrapyrin on variable N-rate fertilization. In “Proceedings of the Sec-
ond International Conference on Site Specific Management for Agricultural Systems, Blooming-
tonlhlinneapolis, MN, 27-30 March 1994” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 571-576. ASA, CSSA, and SSSA, Madison, WI.
Malzer, G. L., Copeland, P. J., Davis, J. G., Lamb, J. A., Robert, P. C., and Bruulsema, T. W. (1996).
Spatial variability of profitability in site-specific N management. In “Proceedings of the Third In-
ternational Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C.
Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 967-975. ASA,
CSSA, and SSSA, Madison, WI.
Mausbach, M. J., and Wilding, L. P. (Eds.) (1991). “Spatial Variabilities of Soilsand Landforms,” SSSA
Special Publ. No. 28. SSSA, Madison, WI.
Mausbach, M. J., Lytle. D. J., and Spivey, L. D. (1993). Application of soil survey information to soil
specific farming. In “Proceedings of Soil Specific Crop Management: A Workshop on Research
and Development Issues, 14-16 April 1992, Minneapolis, MN” (P. C. Robert, R. H. Rust, and
W. E. Larson, Eds.), pp 57-68. ASA, CSSA, and SSSA, Madison, WI.
McBratney, A. B., Whelan, B. M., and Viscarra Rossel, R. A. (1996). Spatial prediction for precision
agriculture. In “Proceedings of the Thiid International Conference on Precision Agriculture, Min-
neapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Mis-
cellaneous Publication, pp. 331-342. ASA, CSSA, and SSSA, Madison, WI.
McBratney, A. B.. Whelan, B. M., and Sham, T. M. (1997). Variability and uncertainty in spatial, tem-
poral, and spatiotemporal crop-field and related data. In “Precision Agriculture: Spatial and Tem-
poral Variability of Environmental Quality” (J. V. Lake, G. R. Bock, and J. A. Goode, Eds.), pp.
141-160. Wiley, New York.
McCann, I. R., and Stark, J. C. (1993). Method and apparatus for variable application of irrigation wa-
ter and chemicals, U.S. Patent No. 5,246,164, September 21.
McCollum, R. E. (1991). Buildup and decline of soil phosphorus: 30-Year trends on a typic umprabu-
ult. Agron. J. 83,77-85.
McKenzie, R. C., Sprout, C. H., and Clark, N. F. (1983). The relationship of the yield of irrigated bar-
ley to soil salinity as measured by several methods. Can. J. Soil Sci. 63,519-218.
McLean, E. 0. (1982). Soil pH and lime requirement. In “Methods of Soil Analysis: Part 2-Chemi-
cal and Microbiological Properties” (A. L. Page, R. H. Miller, and D. R. Keeney, Eds.), 2nd ed.,
pp. 199-224. ASA and SSSA, Madison, WI.
McLean, E. 0..and Brown, J. R. (1984). Crop response to lime in the midwestern United States. In
“Soil Acidity and Liming” (F. Adams, Ed.), 2nd ed., pp. 267-303. ASA, CSSA, and SSSA, Madi-
son, WI.
Meisinger, J. J., and Randall, G. W. (1991). Estimating nitrogen budgets for soil-crop systems. In
“Managing Nitrogen for Groundwater Quality and Farm Profitability” (R. F. Follett, D. R. Keeney,
and R. M. Cruse, Eds.),pp. 85-124. SSSA, Madison, WI.
Meisinger, J. J., Magdoff, R.R., and Schepers, J. S. (1992). Predicting N fertilizer needs for corn in hu-
mid regions: Underlying principles. In “Predicting N Fertilizer Needs for Corn in Humid Regions”
78 FRANCISJ. PIERCE AND PETER NOWAK

(B. R. Bock and K. R. Kelly, Eds.), Bull. No. Y-226, TVAlNFERC-92I2. pp. 6-27. National Fer-
tilizer and Environmental Research Center, Tennessee Valley Authority, Muscle Shoals, AL.
Miller, M. P., Singer, M. J., and Nielsen, D. R. (1988). Spatial variability of wheat yield and soil prop-
erties on complex hills. Soil Sci. SOC.Am. J. 52, 1133-1141.
Moore, G. (1997). Moore’s law, hnp://ww.inrel.com
Moore, I. D., Gessler, P. E., Neilsen, G. A., and Peterson, G. A. (1993). Soil attribute prediction using
terrain analysis. Soil Sci. SOC.A m J. 57,443-452.
Moran, M. S., Inoue, Y., and Barnes,E. M. (1997). Opportunities and limitations for image-based re-
mote sensing in precision crop management. Remote Sensing and the Environ. 61,319-346.
Mortensen, D. A., Johnson, G. A., and Young, L. J. (1993). Weed distribution in agricultural fields. In
“Proceedings of Soil Specific Crop Management: A Workshop on Research and Development Is-
sues, 14-16 April 1992, Minneapolis, MN” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
pp. 113-124. ASA, CSSA, and SSSA, Madison, WI.
Mortensen, D. A., Johnson, G. A., Wyse, D. Y.,and Martin, A. R. (1995). Managing spatially variable
weed populations. In “Proceedings of the Second International Conference on Site Specific Man-
agement for Agricultural Systems, BloomingtonlMinneapolis,MN, 27-30 March 1994” (P. C.
Robert, R. H. Rust, and W. E. Larson, Eds.), ASAMiscellaneous Publication. pp. 397-415. ASA,
CSSA, and SSSA. Madison, WI.
Mortensen, D. A., Anita Dieleman. J., and Johnson, G. A. (1998). Weed spatial variation and weed man-
agement. In “Integrated Weed and Soil Management” (J. L. Hatfield, D. D. Buhler, and B. A. Stew-
art, Eds.), pp. 293-309. Sleeping Bear Press, Chelsea, MI.
Mueller, D. K., Hamilton, P. A.. Helsel, D. R.. Hitt, K. J., and Ruddy, B. C. (1995). Nutrients in ground
water and surface water of the United States-An analysis of data through 1992, Water-Resources
Investigations Report No. 95-403 I . United States Geological Survey, Denver.
Mulla, D. J. (1997). Geostatistics, remote sensing, and precision farming. In “Precision Agriculture:
Spatial and Temporal Variability of Environmental Quality” (J. V. Lake, G. R. Bock, and J. A.
Goode, Eds.), pp. 100-1 15. Wiley, New York.
Mulla, D. J., and Schepers, J. S. (1997). Key processes and properties for site-specific management. In
“The State of Site-Specific Management for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA
Miscellaneous Publication, pp. 1-18. ASA, CSSA, and SSSA, Madison, WI.
Mulla, D. J., Bhatti, A. U., and Kunkel, R. (1990). Methods for removing spatial variability from field
research trials. Adv. Soil Sci. 13,201-213.
Mulla, D. J., Bhatti, A. U., Hammond, M. W., and Benson, J. A. (1992). A comparison of winter wheat
yield and quality under uniform versus spatially variable fertilizer management. Agric. Ecosys-
tems Environ. 38,301 -3 1I .
Mulla, D. J., Perillo, C. A., and Cogger, C. G. (1996). A site-specific farm-scale GIS approach for re-
ducing groundwater contamination by pesticides. J. Environ. Qualiry 25,419-425.
Nafziger. E. D. (1994). Corn planting data and plant population. J. Prod. Agric. 7,59-62.
Nafziger, E. D. (1996). Effects of missing and two-plant hills on com grain yield. J. Prod. Agric. 9,
238-240.
National Academy of Public Administration (NAPA) (1995). “The Global Positioning System: Chart-
ing the Future.’’ NAPA, National Academy Press, Washington, DC.
National Aeronautics and Space Administration (NASA) (1998). “The Remote Sensing Tutorial: An
Online Handbook. NASA’s Goddard Space Flight Center. hrtp://code935.gsfc.nusa.gov/Tutori-
al/Stan.hrml.
National Agricultural Statistics Service (1997). Farm computer usage. In “Agriculture Across Michi-
gan,” Vol. 18, No. 9, p. 1. Michigan Agricultural Statistics Service, Lansing, MI.
National Research Council (NRC) (1997). “Precision Agriculture in the 21st Century: Geospatial and
Information Technologies in Crop Management.” NRC National Academy of Sciences, Wash-
ington, DC.
ASPECTS OF PRECISIONAGRICULTURE 79

Nelson, T. (1993). Sprayvision, a selective sprayer technology developed in North America. Proc.
Weed Sci. Suc. Amer: 33,43.
Nielsen, R. L. (1995). Planting speed effects on stand establishment and grain yield of corn. J. Pmd.
Agric. 8,391-393.
Nowak, P. (1997).A sociological analysis of site specific management. In “The State of Site-Specific
Management for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA Miscellaneous Publica-
tion, pp. 397-422. ASA, CSSA, and SSSA, Madison, WI.
Nowak, P., and Korsching, P. F. (1998).The human dimensions of soil and water conservation: A his-
torical and methodological perspective. In “Advances in Soil and Water Conservation” (F. J.
Pierce and W. W. Frye, Ms.), pp. 219-231. Sleeping Bear Press, Chelsea, MI.
Nowak, P., Shepard. R., and Madison, F. (1998). Farmers and manure management: A critical analy-
sis. In “Animal Waste Utilization: Effective Use of Manure as a Soil Resource” (J. L. Hatfield and
B. A. Stewart, Eds.), pp. 1-32. Ann Arbor Press, Chelsea, MI.
Nuspl, S. J.. Randolf, W. W., and Guthland, R. (1996). Use of injection for site-specific chemical ap-
plication. In “Proceedings of the Third International Conference on Precision Agriculture, Min-
neapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.). ASA Mis-
cellaneous Publication, pp. 739-744. ASA, CSSA, and SSSA, Madison, WI.
O’Conner, M.. Bell, T.. Elkaim, G., and Parkinson, B. (1996).Automatic steering of farm vehicles us-
ing GPS. In “Proceedings of the Thud International Conference on Precision Agriculture, Min-
neapolis, MN, 23-26 June 1996” (F! C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Mis-
cellaneous Publication, pp. 767-777. ASA, CSSA, and SSSA, Madison, WI.
Odeh, I. 0. A., McBratney, A. B., and Chittleborough, D. J. (1995). Further results on spatial predic-
tion of soil properties from landform attributes derived from digital elevation models: Hetertopic
cokriging and regression-kriging models. Ceudennn 67,215-226.
Olieslagers. R., Ramon, H., and De Baerdemaeker, J. (1996). Design of a centrifugal spread for site-
specific fertilizer application. In “Proceedings of the Third International Conference on Precision
Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson,
a s . ) ,ASA Miscellaneous Publication, pp. 745-756. ASA, CSSA, and SSSA. Madison, WI.
Olson, R. (1995). How a crop consulting firm is using precision agriculture technology. In “Proceed-
ings of the Second International Conference on Precision Agriculture, Minneapolis, MN, 27-30
March 1994” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication,
pp. 755ff. ASA, CSSA. and SSSA, Madison, WI.
Olson, R. A,, and Kurtz, L. T. (1982).Crop nitrogen requirements, utilization, and fertilization. In “Ni-
trogen in Agricultural Soils” (F. J. Stevenson, Ed.), Agronomy Monograph No. 22, pp. 567-604.
ASA. CSSA, and SSSA, Madison, WI.
Olson, R. A., Anderson, R. N., Frank, K. D., Grabouski, P. H., Rehm, G. W., and Shapiro, C. A. (1987).
Soil testing interpretations: Sufficiency vs. buildup and maintenance. In “Soil Testing: Sampling,
Correlation, Calibration, and Interpretation” (J. R. Brown, Ed.), SSSA Special Publ. No. 21, pp.
41-52. SSSA, Madison, WI.
Ortmann, G. F., Patrick, G. F., and Musser, W. N. (1994). Use and rating of computers by large-scale
U S . cornbelt farmers. Cumpurers Electronics Agric. 10,31-43.
Page, N. R. (1974).Estimation of organic matter in Atlantic coastal plain soils with a color-difference
meter. Agmn. J. 66,652-653.
Palmer, R. (1991). Progress report of a local positioning system. In “Automated Agriculture for the
21st Century. Proceedings of the 1991 Symposium, Chicago, IL, 16-17 December,”ASAE Publ.
No. 11-91, pp. 402-408. ASAE, St. Joseph, MI.
Palmer, R. (1995).Positioning aspects of site-specific applications. In “Proceedings of the Second In-
ternational Conference on Site Specific Management for Agricultural Systems, Bloomingtonl
Minneapolis, MN, 27-30 March 1994” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA
Miscellaneous Publication, pp. 613-618. ASA, CSSA, and SSSA, Madison, WI.
80 FRANCIS J. PIERCE AND PETER NOWAK

Pan, W. L., Huggins, D. R., Malzer, G. L., Douglas, C. L.,Jr., and Smith, J. L. (1997). Field hetero-
geneity in soil-plant nitrogen relationships: Implications for site-specific management. In “The
State of Site-SpecificManagementfor Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA Mis-
cellaneous Publication, pp. 81-99. ASA. CSSA, and SSSA, Madison, WI.
Peck, A. J. ( 1983). Field variability of soil physical properties. Adv. Im’g.2, 189-22 1.
Peck, T. R., and Melsted, S. W. (1973).Field sampling for soil testing. In “Soil Testing and PlantAnaly-
sis” (L. M. Walsh and J. D. Beacon, Eds.), pp. 67-75. SSSA, Madison, WI.
Pedigo, L. P. (1994). Introduction to sampling arthropod populations. In “Handbook for Sampling
Methods for Arthropods in Agriculture” (L. P. Pedigo and G. D. Buntin, Eds.), pp. 1-11. CRC
Press, Boca Raton, FL.
Perry, G. (1998. February). A RIN Report: The News About GLOSNASS. GPS World, 44-45.
Peterson, T. A. (1997). Setting up side X side comparisons with a yield monitor. In “Crop Insights,”
Vol. 7, No. 10. Pioneer Hi-Bred International, Des Moines, IA.
Peterson, T. A., and Beck, R. H. (1997). Site-specific management: Technology transfer and educa-
tional needs. In ‘The State of Site-Specific Management for Agriculture” (F. J. Pierce and E. J.
Sadler, Eds.), ASA Miscellaneous Publication, pp. 423-430. ASA, CSSA, and SSSA. Madison,
WI.
Pierce, F. J. (1995, November). If (condition) then (action). ug/INNOVATOR $4.
Pierce, F. J. (1997a. October). SSM is TID.ug/INNOVATOR 5,s.
Pierce, F. J. (1997b. June). Precision farming: Good for agriculture, good for the environment? ug/IN-
NOVATOR 4 5 .
Pierce, F. J., and Gilliland, D. C. (1997). Soil quality control. In “Soil Quality for Crop Production Sys-
tems and Ecosystem Health” (E. G. Gregorich and M. R. Carter, Eds.), Developments in Soil Sci-
ence Vol. 25, pp. 203-220. Elsevier, Amsterdam.
Pierce, F. J., and Lal, R. (1991). Soil management in the 21st century. In “Soil Management for Sus-
tainability” (R. La1 and F. J. Pierce, Eds.), pp. 175-180. Soil and Water Conservation Society,
Ankeny, IA.
Pierce, F. J., and Sadler, E. J. (Eds.)(1997). “The State of Site-SpecificManagement for Agriculture.”
ASA Miscellaneous Publication. ASA. CSSA, and SSSA. Madison, WI.
Pierce, F. J., Robert, P. C., and Mangold, G. (1994). Site-specificmanagement: The pros, the cons, and
the realities. In “Proceedingsof the International Crop Management Conference, Iowa State Uni-
versity,” pp. 17-21. Iowa State Univ. Press, Ames.
Pierce, F. J., Warncke, D. D., and Everett, M. W. (1995). Yield and nutrient availability in glacial soils
of Michigan. In “Proceedings of the Second International Conference on Site Specific Manage-
ment for Agricultural Systems, BloomingtonlMinneapolis, MN, 27-30 March 1994” (P. C.
Robert, R. H. Rust, and W. E. Larson, Eds.),ASA Miscellaneous Publication, pp. 133-151. ASA,
CSSA, and SSSA, Madison, WI.
Pierce, F. I., Anderson, N., Colvin, T. S.. Schueller. J. K.,Humburg, D., and McLaughlin, N. (1997).
Yield mapping. In ‘The State of Site-Specific Management for Agriculture” (F. J. Pierce and
E. J. Sadler, Eds.),ASAMiscellaneousPublication, pp. 203-236. ASA. CSSA, and SSSA, Madi-
son, WI.
Pioneer Hi-Bred International,Inc. (1997). Narrow-row corn, research brochure. Pioneer Hi-Bred, Des
Moines, IA.
Pointon, J. (1997. July). A differential of opinion. GPS World, 12.
Porter, P. M., Hicks, D.R., Lueschen, W. E., Ford, J. H., Warnes, D. D., and H0verstad.T. R. (1997). Corn
response to row width and plant population in the northern corn belt. J. Pmd. Agric. 10,293-300.
Prairie Agricultural Machinery Institute (PAMI) (1987). ’Qe Series V1114-5360. No Till Drill. Evalu-
ation report 519. PAMI, Humbolt, SK,Canada.
Price, R. R., and Gaultney, L. D. (1993). Soil moisture sensor for predicting seed planting depth. Tmns.
Am SOC.Agric. Eng. 36,1703-1711.
ASPECTS OF PRECISION AGRICULTURE 81

Rasmussen, P. E., and Rhcde, C. R. (1991). Tillage, soil depth and precipitation effects on wheat re-
sponse to nitrogen. Soil Sci. SOC.Am. J. 55, 121-1 24.
Redulla, C. A.. Havlin, J. L., Kluitenberg, G. J., Zhang, N., and Shrock, M. D. (1996). Variable nitro-
gen management for improving groundwater quality. In “Proceedings of the Third International
Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1 9 9 6 (P. C. Robert, R. H.
Rust, and W. E. Larson, Eds.),ASA MiscellaneousPublication,pp. 1101-1 110. ASA, CSSA, and
SSSA. Madison, WI.
Reichenberger, L. (1996, mid-February). A vote for variable seeding. Farm J. 120, 18D.
Reichenberger, L., and Russnogle, J. (1989, mid-March). Farm by the foot. Farm J. 113, 11-15.
Robert, P. C., Rust, R. H.. and Larson, W. E. (Eds.) (1993). “Proceedings of Soil Specific Crop Man-
agement: A Workshop on Research and Development Issues, 14-16 April 1992, Minneapolis,
MN.” ASA, CSSA, and SSSA, Madison, WI.
Robert, P. C., Rust, R. H., and Larson, W. E. (Eds.) (1995). “Proceedings of the Second International
Conference on Site Specific Management for Agricultural Systems, BloomingtonlMinneapolis.
MN, 27-30 March 1994,” ASA Miscellaneous Publication. ASA, CSSA, and SSSA. Madison,
WI.
Robert, P. C., Rust, R. H., and Larson, W. E. (Eds.) (1996). “Proceedings of the Third International
Conferenceon Precision Agriculture, Minneapolis,MN, 23-26 June 1996,” ASA Miscellaneous
Publication. ASA, CSSA, and SSSA, Madison, WI.
Rogers, E. (1995). “Diffusion of Innovations,” 4th ed. Free Press, New York.
Rossi, R. E.. Mulla, D. J., Journal,A. G., and Franz, E. H. (1992). Geostatisticaltools for modeling and
interpretingecological spatial dependence. Ecol. Monogr: 62,277-3 14.
Sadjadi, F. (1996).Application of image understandingtechnology in precision agriculture: Weed clas-
sification and crop row guidance. In “Proceedings of the Third International Conference on Pre-
cision Agriculture, Minneapolis,MN, 23-26 June 1996“ (P. C. Robert, R. H. Rust, and W. E. Lar-
son, Eds.), ASA Miscellaneous Publication,pp. 779-784. ASA, CSSA, and SSSA, Madison, WI.
Sadler, E. J., and Russell, G. (1997). Modeling crop yield for site-specific management. In “The State
of Site-Specific Management for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA Miscella-
neous Publication,pp. 69-79. ASA, CSSA, and SSSA, Madison, WI.
Sadler. E. J., Camp, C. R., Evans, D. E., and Usrey, L. J. (1996). A site-specific center pivot irrigation
system for highly-variable coastal plain soils. In “Proceedings of the Third International Confer-
ence on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P.C. Robert, R. H. Rust,
and W. E. Larson. Eds.), ASAMiscellaneous Publication, pp. 827-834. ASA, CSSA, and SSSA,
Madison, WI.
Sadler, E. J.. Busscher, W. J., Bauer, P. J., and Karlen, D. L. (1998). Spatial scale requirements for pre-
cision farming: A case study in the southeasternUSA. Agmn. J. 90,191-197.
Saunders,W. P., Larscheid, G., Blackmore, B. S., and Stafford, J. V. (1996). A method for direct com-
parison of differential global positioning systems suitable for precision farming. In “Proceedings
of the Third International Conference on Precision Agriculture, Minneapolis, MN, 23-26 June
1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp.
663-674. ASA, CSSA, and SSSA, Madison, WI.
Sawyer, J. E. (1994). Concepts of variable rate technology with considerations for fertilizer applica-
tion. J. Prod. Agric. 7, 195-201.
Schepers, J. S., Francis, D. D., Vigil, M., and Below, F. E. (1992). Comparison of corn leaf N concen-
tration and chlorophyll meter readings. Commun. Soil Sci. PIanr Anal. 23,2173-2187.
Schepers, J. S., Blackmer. T. M., Shah, T., and Christensen, N. (1996). Remote sensing tools for site-
specific management. In “Proceedings of the Third International Conference on Precision Agri-
culture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 315-319. ASA, CSSA, and SSSA, Madison, WI.
Schnitkey. G., Hopkins, J., and Tweeten, L. (1996). An economic evaluation of precision fertilizer ap-
82 FRANCISJ. PIERCE AND PETER NOWAK

plications on corn-soybean fields. In “Proceedings of the Third International Conference on Pre-


cision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Lar-
son. Eds.). ASA Miscellaneous Publication, pp. 977-987. ASA, CSSA, and SSSA, Madison, WI.
Schueller, J. K. (1991). In-field site-specific crop production. In “Automated Agriculture for the 21st
Century. Proceedings of the 1991 Symposium, Chicago, IL,16-17 December.” ASAE Publ. No.
11-91, pp. 291-292. ASAE, St. Joseph, MI.
Schueller. J. K. (1992). A review and integrating analysis of spatially-variable control of crop produc-
tion. Fertilizer Res. 33, 1-34.
Scopel. A. L., Ballare, C. L., and Rodosevich, S. R. (1994). Photosimulation of seed germination dur-
ing soil tillage. New Phyfol. 126, 145-152.
Scorer, A. G. (1991). The development of datatrack. J. Navigation 44,37-47.
Sharpley, A. N. (1997). Dispelling common myths about phosphorus in agriculture and the environ-
ment, Watershed Science Institute Technical Paper. United States Department of Agriculture-Nat-
ural Resources Conservation Service, Washington, DC.
Shearer, S. A., and Jones, P. T. (1991). Selective application of post-emergence herbicides using pho-
toelectrics. Trans. Am. SOC.Agric. Eng. 34, 1661-1666.
Shopshire, G. J.. VonBargen, K., and Mortensen, D. A. (1990). Optical reflectance sensor for detect-
ing plants. In “Optics in Agriculture” (J. A. DeShazer and G. E. Meyers, Eds.), pp. 225-235. So-
ciety of Photo-Optical Instrumental Engineering, Bellingham, WA.
Shumway, R. H. (1988). “Applied Statistical Time Series Analysis.” Prentice-Hall, Englewood Cliffs, NJ.
Smith, S. A., Essington, M. E., Howard, D. D., Tyler, D. D., and Wilkerson, J. (1998). Site-specific nu-
trient management: Variability in cotton yield response and soil chemical characteristics. Better
Crops 82,12-14.
Soil Survey Division Staff (1993). Soil survey manual, United States Department ofAgriculture, Hand-
book No. 18. U.S. Government Printing Office, Washington, DC.
Sojka, R. E., Karlen, D. L., and Sadler, E. J. (1988). Planting geometries and the efficient use of water
and nutrients. In “Cropping Strategies for Efficient Use of Water and Nitrogen” (W. L. Hargrove,
Ed.), ASA Special Publ. No. 51, pp. 43-68. ASA, CSSA, and SSSA, Madison, WI.
Sokal, R. R., and Rohlf, F. J. (1995). “Biometry: The Principles and Practice of Statistics in Biologi-
cal Research.” 3rd ed. Freeman, New York.
Solohub, M. P., van Kessel, C.. and Pennock, D. J. (1996). The feasibility of variable rate N fertiliza-
tion in Saskatchewan. In “Proceedings of the Third International Conference on Precision Agri-
culture, Minneapolis, MN, 23-26 June 1996“ (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 65-73. ASA, CSSA, and SSSA, Madison, WI.
Spomer, R. G., and Piest, R. F. (1982). Soil productivity and erosion of Iowa loess soils. Trans. Am.
SOC.Agric. Eng. 25, 1295- 1299.
Stafford,J. V. (Ed.) (1 996a). Spatially variable field operations [Special issue]. Computers Electronics
Agric. 14,99-253.
Stafford, J. V. (1996b). Essential technology for precision agriculture. In “Proceedings of the Third In-
ternational Conference on Precision Agriculture. Minneapolis, MN, 23-26 June 1996” (P. C.
Robert, R. H. Rust, and W. E. Larson, Eds.). ASA Miscellaneous Publication, pp. 595-604. ASA,
CSSA, and SSSA, Madison, WI.
Stafford,J. V., and Miller. P. C. H. (1993). Spatially selective application of herbicides to cereal crops.
Computers Electronics Agric. 9,217-229.
Stafford, J. V., and Miller, P. C. H. (1996). Spatially variable treatment of weed patches. In “Proceed-
ings of the Third International Conference on Precision Agriculture, Minneapolis, MN, 23-26
June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Us.),ASA Miscellaneous Publication,
pp. 465-474. ASA, CSSA, and SSSA, Madison, WI.
Stafford, J. V., LeBars, J. M.. and Ambler, B. (1996). A hand-held data logger with integral GPS for
producing weed maps by field walking. Computers Electronics Agric. 14,235-247.
ASPECTS OF PRECISION AGRICULTURE 83

Stein, A., Hoosbeek. M. R., and Sterk. G . (1997). Space-time statistics for decision support to smart
farming. In “Precision Agriculture: Spatial and Temporal Variability of Environmental Quality”
(J. V. Lake, G. R. Bock, and J. A. Goode, Eds.), pp. 120-130. Wiley, New York.
Stevens, M. D. (1993). Satellite remote sensing for agricultural management: Opportunities and logis-
tic constraints. J. Phorogrammeiry Remore Sensing 48,29-34.
Stevenson, C., and van Kessel, C. (1996). A landscape-scale assessment of the nitrogen and non-ni-
trogen benefits of pea in a crop rotation. In “Proceedings of the Third International Conference on
Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R.H. Rust, and W. E.
Larson, Eds.), ASA Miscellaneous Publication, pp. 55-63. ASA, CSSA, and SSSA, Madison, Wl.
Stevenson, F. J. (1982). Organic forms of soil nitrogen. In “Nitrogen in Agricultural Soils” (F. J. Steven-
son, Ed.), Agronomy Monograph No. 22, pp. 67-122. ASA, CSSA, and SSSA, Madison, WI.
Steward, B. L. (1994). Modeling and simulation of a chemical injection system, M.S. thesis. South
Dakota State University, Brookings.
Stone, J. R., Gilliam, J. W., Cassel, D. K., Daniels, R. B., Nelson, L. A., and Kleiss, H.J. (1985). Ef-
fects of erosion and landscape position on the productivity of piedmont soils. Soil Sci. SOC.Am.
J. 49,987-991.
Stone, M. (1991). Control system applications. In “Automated Agriculture for the 21st Century. Pro-
ceedings of the 1991 Symposium. Chicago, IL, 16-17 December,” ASAE h b l . No. 11-91, pp.
163-165.ASAE. St. Joseph, MI.
Stone, M. L., Solie, J. B., Raun, W. R., Whitney, R. W., Taylor, S. L., and Ringer, J. D. (1996). Use of
spectral radiance for correcting in-season fertilizer nitrogen deficiencies in winter wheat. Trans.
Am. Soc. Agric. Eng. 39,1623- 1631.
Sudduth, K. A. (1998). Engineering and application of precision farming technology. In “Integrated
Weed and Soil Management” (J. L. Hatfield, D. D. Buhler, and B. A. Stewart, Eds.), pp. 31 1-33 1.
Sleeping Bear Press, Chelsea, MI.
Sudduth, K. A,, Hummel, J. W., and Birrell, S. J. (1997). Sensors for site-specific management. In “The
State of Site-Specific Management for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA Mis-
cellaneous Publication, pp. 69-79. ASA, CSSA, and SSSA, Madison, WI.
Swinton, S. M., and Ahmad, M. (1996). Returns to farmer’s investments in precision agriculture equip-
ment and services. In “Proceedings of the Third International Conference on Precision Agricul-
ture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.),
ASA Miscellaneous Publication, pp. 1009-1018. ASA, CSSA, and SSSA, Madison, WI.
Taylor, J., and Wacker. W. (1997). “The 500 Year Delta: What Happens after What Comes Next.”
HarperBusiness, New York.
Taylor, L. R. (1984). Assessing and interpreting the spatial distributions of insect populations. Annu.
Rev. Enromol. 29,32 1-357.
Tevis, J. W., Whittaker, A. D., and McCauley, D. J. (1991). Efficient use of data in the kriging of soil
pH, ASAE Paper No. 9 1-7047. ASAE, St. Joseph, MI.
Thomison, P. R., and Jordan, D. M. (1995). Plant population effects on corn hybrids differing in ear
growth habit and prolificacy. J. Prod. Agric. 8,394-400.
Tollenaar, M. (1991). Physiological basis of genetic improvement of maize hybrids in Ontario from
1959 to 1988. Crop Sci. 31, 119-124.
Tomer, M. D., Anderson, J. L., and Lamb, J. A. (1995). Landscape analysis of soil and crop data using
regression. In “Proceedings of the Second International Conference on Site Specific Management
for Agricultural Systems, BlwmingtonlMinneapolis, MN, 27-30 March 1994” (P. C. Robert,
R. H. Rust, and W. E. Larson, Eds.), ASA Miscellaneous Publication, pp. 273-284. ASA, CSSA,
and SSSA, Madison, WI.
Trangmar, B. B., Yost, R. S., and Uehara, G. (1985). Application of geostatistics to spatial studies of
soil properties. Adv. Agron. 38,45-94.
’Qler. D. A., Roberts, D. W., and Nielsen, G. A. (1997). Location and guidance for site-specific man-
84 FRANCIS J. PIERCE AND PETER NOW=

agement. In “The State of Site-Specific Management for Agriculture” (F. J. Pierce and E. J. Sadler,
Eds.), ASAMiscellaneous Publication, pp. 161-181. ASA, CSSA, and SSSA, Madison, WI.
U S . Department of Agriculture, Economic Research Service (USDA-ERS) (1987). “Farm Drainage
in the United States: History, Status, and Prospects” (George A. Pavelis, Ed.), Publ. No. 1455.
USDA-ERS, Washington, DC.
U.S. Department of Agriculture, Economic Research Service (USDA-ERS) (1997). “Agricultural Re-
sources and Environmental Indicators, 1996-97,”Agricultural Handbook No. 712. USDA-ERS,
Natural Resources and Environment Division, Washington, DC.
Verhagen, J., and Bouma, J. (1997).Modeling soil variability. In “The State of Site-Specific Manage-
ment for Agriculture” (F. J. Pierce and E. J. Sadler, Eds.), ASA Miscellaneous Publication, pp.
55-67. ASA, CSSA, and SSSA, Madison, WI.
Verhagen, A.. Booltink, H. W. G., and Bouma, J. (1995a). Site-specific management: Balancing pro-
duction and environmental requirements at farm level. Agric. Sysrems 49,369-384.
Verhagen, A., Verburg, P., Sybesma, M., and Bouma, J. (1995b).Terrain modeling as a basis for opti-
mal agroecological land management using dynamic simulation. In “Proceedings of the Second
International Conference on Site Specific Management for Agricultural Systems, Bloomington/
Minneapolis, MN, 27-30 March 1994” (P. C. Robert, R. H. Rust, and W. E. Larson, Eds.), ASA
Miscellaneous Publication, pp. 229-250. ASA, CSSA, and SSSA, Madison, WI.
Vetsch, J. A., Malzer, G. L., Robert, P. C., and Huggins, D. R. (1995). Nitrogen specific management
by soil condition: Managing fertilizer nitrogen in corn. In “Proceedings of the Second Interna-
tional Conference on Site Specific Management for Agricultural Systems, Bloomington/Min-
neapolis, MN, 27-30 March 1994” (P. C. Robert. R. H. Rust, and W. E. Larson, Eds.),ASA Mis-
cellaneous Publication, pp. 465-473. ASA, CSSA, and SSSA, Madison, WI.
Vetsch, W. C., Duke, H. R., and Iremonger, C. J. (1996).Assessment of plant nitrogen in imgated corn.
In “Proceedings of the Third International Conference on Precision Agriculture, Minneapolis,
MN, 23-26 June 1996” (R C. Robert, R. H. Rust, and W. E. Larson. Eds.), ASA Miscellaneous
Publication, pp. 23-32. ASA, CSSA, and SSSA, Madison, WI.
Vitosh, M. L., Johnson, J. W., and Mengel, D. B. (1996).Tri-state fertilizer recommendations for corn,
soybeans, wheat and alfalfa, Extension Bull. No. E-2567.Michigan State University, East Lans-
ing.
Von Groenendael, J. M. (1988).Patchy distribution of weeds and some implications for modeling pop-
ulation dynamics: A short literature review. Weed Res. 28,437-441.
Voorhees, W. B., Allmaras, R. R., and Lindstrom, M. J. (1993).Tillage considerations in managing soil
variability. In “Proceedings of Soil Specific Crop Management: A Workshop on Research and De-
velopment Issues, 14-16 April 1992, Minneapolis, MN” (P. C. Robert, R. H.Rust, and w. E. Lar-
son, Eds.),pp. 95-111. ASA, CSSA. and SSSA, Madison, WI.
Walker, B. D., Haugen-Kozyra, K.,and Wang. C. (1996).Effects of long-term cultivation on a morainal
landscape in Alberta, Canada. In “Proceedings of the Third International Conference on Precision
Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C. Robert, R. H. Rust, and W. E. Larson.
Eds.), ASA Miscellaneous Publication, pp. 107-116. ASA, CSSA, and SSSA, Madison, WI.
Walker, P. H., Hall, F. F., and Protz, R. (1968). Relation between landform parameters and soil prop-
erties. Soil Sci. SOC.Am. Pmc. 32, 101-104.
Wall, R. W., King, B. A., and McCann, I. R. (1996). Center-pivot irrigation system control and data
communications network for real-time variable water application. In “Proceedings of the Third
International Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996” (P. C.
Robert, R. H.Rust, and W. E. Larson, Eds.), ASAMiscellaneous Publication, pp. 757-766. ASA,
CSSA, and SSSA. Madison, WI.
Walters, C. S., Frazier, B. E., Miller, B. C., and Pan, W. L. (1996). Remotely sensed organic matter
mapping for site-specific N management in dryland winter wheat. In “Proceedings of the Third
International Conference on Precision Agriculture, Minneapolis, MN, 23-26 June 1996 (P. C.
ASPECTS OF PRECISION AGRICULTURE 85

Robert, R. H. Rust, and W. E. Larson, Eds.), ASAMiscellaneousPublication, p. 579. ASA, CSSA,


and SSSA, Madison, WI.
Wanick, A. W., Myers, D. E., and Nielsen, D. R. (1986). “GeostatisticalMethods Applied to Soil Sci-
ence. Methods of Soil Analysis. Part I,” 2nd ed., Agronomy Monograph No. 9. ASA, CSSA, and
SSSA, Madison, WI.
Weatherly, E. T., and Bowers, C. G.,Jr. (1997). Automatic depth control of a seed planter based on soil
drying front sensing. Trans. Am. SOC.Agric. Eng. 40,295-305.
Webster, R., and Oliver, M. A. (1990). “Statistical Methods in Soil and Land Resource Survey.” Ox-
ford Univ. Press, New York.
Weibel, R., and Heller, M. (1991). Digital terrain modeling. In “GeographicInformation Systems, Prin-
ciples and Applications” (M. F. Goodchild and D. Rhines, Eds.), pp. 269-297. Taylor & Francis,
New York.
Wesseling, J. (1974). Drainage and crop production. In “Drainage for Agriculture” (J. Van Schilf-
gaarde, Ed.), pp. 7-37. Agronomy Vol. 17. ASA, Madison, WI.
Wibawa, W. D., Dludlu, D. L., Swenson, L. J., Hopkins, D. G., and Dahnke, W. C. (1993). Variable
fertilizer application based on yield goal, soil fertility, and soil map unit. J. Prod. Agric. 6,255-
261.
Woebbecke, D. M., Meyer, G. E., Von Bargen, K.,and Mortensen, D. A. (1995). Shape features for
identifying young weeds using image analysis. Trans. Am SOC.Agric. Eng. 38,271-281.
Wolf, S. A. (1998). Introduction. In “Privatization of Information and Agricultural Industrialization”
(S. A. Wolf, Ed.), pp. xvii-xxv. CRC Press, Boca Raton, FL.
Wolf, S. A., and Nowak, P. (1995). Site-specificity in agriculture: Hard or soft paths for Wisconsin’s
agrichemical dealers. In “Proceedings of the 1995 Fertilizer,Aglime, and Pest Management Con-
ference, University of Wisconsin-Madison.” pp. 121-129. Department of Soil Science, Univer-
sity of Wisconsin-Madison.
Wollenhaupt, N. C.. Wolkowski, R. P., and Clayton, M. K. (1994). Mapping soil test phosphorus and
potassium for variable-rate fertilizer application. J. Prod. Agric. 7,441-448.
Wollenhaupt, N. C., Mulla. D. J., Gotway, C. A., and Crawford, L. A. (1997). Soil sampling and inter-
polation techniques for mapping spatial variability of soil properties. In ‘The State of Site-Spe-
cific Management for Agriculture” (p. J. Pierce and E.J. Sadler, Eds.), ASA Miscellaneous Pub-
lication, pp. 19-53. ASA, CSSA, and SSSA, Madison, WI.
Young, J. L., and Aldag, R. W. (1982). Inorganic forms of nitrogen in soil. In “Nitrogen in Agricultur-
al Soils’’ (F. J. Stevenson, Ed.), Agronomy Monograph No. 22, pp. 43-66. ASA, CSSA, and
SSSA, Madison, WI.

You might also like