You are on page 1of 6

An economic analysis of a weed detection system in

Argentina
Luciana Moltoni1), Andrés Moltoni1), Gerardo Masiá1)
1)
Instituto de Ingeniería Rural, CIA, CNIA, INTA. CC 25, Castelar (1712), Buenos Aires, Argentina. E-mail:
lmoltoni@cnia.inta.gov.ar, amoltoni@cnia.inta.gov.ar, gmasia@cnia.inta.gov.ar

Introduction
The application of herbicides allows producers to have an efficient control of weeds
population. However, due to the ecological impact which herbicide produces, its use should
be minimized. Moreover, a reduction of that amount would represent important economic
savings. Uniform spraying is widely used in herbicide applications in Argentina. On the other
hand, the distribution of weeds is often “patchy”, rather than even or random (Cardina et al.
1995, Gerhards et al. 1997, Wang et al. 2001). Different studies have demonstrated that
weed patches can cover from 80% to as little as a few percent of arable fields (Brown et al.
1990, Thompson et al. 1991, Johnson et al 1995, Rew et al. 1996). Consequently, the
efficiency of weed control can be improved if herbicides are applied only over the weed
infested areas (Wang et al. 2001), reducing the waste which produces the application on
those areas where it is not necessary.
With the aim of reducing ecological impact and costs many approaches have been proposed
for weed detection in the world by using methods of artificial vision (Shearer and Colmes
1990, Woebbecke et al. 1992, Zhang and Chaisattapagon 1995, Tian et al. 1997, 1999). In
other research done by the authors, it was determined that the method of detection of red
and near infrared (RR/NIR) is the most appropriate to be developed in Argentina. With the
application of this technology, savings of 67 million dollars a year would be obtain and
farmer’s profit might be increased in 9% (Moltoni and Moltoni 2005). This system
distinguishes between green plants and soil but it does not make a distinction between
weeds and crop. Despite being only useful for weed control on fallow sites, the amount of
herbicides used in this stage is 29% of the total applied yearly. What is more, no-till is applied
in 60% (16,600,000 ha) of the total area used for agriculture.

Red and near infrared weed detection method (RR/NIR)


RR/NIR detection method is not manufactured in Argentina. However, this imported
equipment can be purchased as individuals modules. The price of each individual module is
1,500 dollars.
In addition, the processing and calibration of the system is done by a computer which is
capable of working with up to sixteen nozzles at the same time and its price is 3,000 dollars.
Taking into account that the length of the boot of a widely adopted sprayer machine in
Argentina is about 25 meters and it has 70 nozzles, the price of the weed detection
equipment reaches an amount of 120,000 dollars (€99,174).
The present study is an economic analysis of this system and its potential use in Argentina.
It was also calculated which should be the appropriate price in order to be widely applied in
our country. A machine equipped with the weed detection system (M2) was compared to a
machine without it (M1) by using classical machinery operating cost analysis and
Mathematica 5.0 software. To determine the effect of this technology on machinery
operating cost, the mean cost was used. The basic parameters employed in order to
calculate the mean cost were: a national manufacture sprayer machine which market price is
around 70,000 dollars (€57,852), estimating the remaining value of the machine as a 30% of
its price, an useful life of 15 years and repair and maintenance charges of 1x10-4 x hour-1.
The yearly interest rate was set in 4%.
It was determined that mean cost would depend on weed coverage and the number of
hectares worked. As the number of hectares worked increases, the mean cost will be lower.
On the other hand, as weed coverage percentage of the field is greater, the mean cost will
increase.

MeC = f (ha, Wc)

Where: MeC Mean Cost


ha hectares
Wc Weed coverage, (0 < Wc < 1)

When M1 is utilized, Wc will be always equal to 1 in view of the fact that herbicide application
is the same as that situation in which there is total weed coverage. Nevertheless, for M2, the
mean cost do depend on weed coverage. As Wc tends to one, the benefits of this technology
will decrease and it will be more appropriate to use M1.
Figure 1(a) shows the behavior of mean cost for both machines. The intersection of the
surfaces represents points where mean cost is equal for both machines. Consequently, it
would be indifference for the farmer to choose between M1 and M2.
On Figure 1(b) it could be seen the intersection curve of both mean cost functions, for M1
and M2, and the projection of that intersection on plane XY, setting two zones. For Wc and
ha combinations situated on the right side of the intersection curve (zone 1), a lower mean
cost would be obtained by using M1. On the other hand, combinations situated on the left of
the intersection curve (zone 2) would obtain a benefit by the use of M2.
Figure 1-(a): Mean cost for M1 and M2 . (b): Intersection curve and its projection on plane
XY. (US dollars)

Figure 2 shows the level curves for weed coverage of 30, 50 and 70% for M2 and the curve
of mean cost for M1. As result of this graphic, the quantity of hectares that the farmer should
work yearly in fallow in order to obtain the benefits of the weed detection system is obtained.

Figure 2: Mean cost of M1 and mean cost of M2 as a function of hectares worked yearly in
fallow for 30, 50 and 70% of weed coverage (US dollars)

For weed coverage of 50%, farmers should work over 3,791 hectares (P2), in order to
perceive the benefits of this new technology. Assuming that of all herbicide applications
done by the sprayer machine yearly, only a quarter of them are assigned for chemical weed
control in fallow, the farmer should worked over 15,200 hectares yearly. Taking this into
account, farmers which work less than 15,200 hectares yearly would not adopt this system
as they would not perceive any benefit of it. This result shows that a large amount of
agricultural producers, specially medium and small ones, are excluded from the usage of this
equipment and, according to its price, the technology would not be extensively adopted.
The same analysis was done for 30 and 70% weed coverage; P1 and P3 respectively
(Figure 2). As the number of hectares worked yearly grows and weed coverage decrease, it
is more appropriate the adoption of the system.
No considerations were made about the consequences on crop yields produced by different
weed coverage before making the control of them. However, this should be taken into
account in future studies.

Setting the price


This technology should have a lower price so as to be widely adopted in our country. This
might be achieved if local industries manufacture the equipment. Although most of the inputs
involved in the manufacture process of this system are imported, this effect is neutralized,
and even inverted by the low wages which local workers are earning.
In order to estimate the price, a range of weed coverage which represents an ordinary
situation was set. Simultaneously, a range of yearly worked hectares was established.
Assuming that weed coverage between 50 and 70% and an annually worked area between
4,000 and 10,000 hectares represent an ordinary situation in Argentina, the results shown in
figure 3 were obtained. This surface represents the points where mean cost for both
machines, with and without the weed detection system, is equal. At the same time, this
shows as a dependent variable the price that the system should have so as the equalization
takes place.
Depending on these assumptions, the price should be between 18,880 (€15,603) (Figure 3,
P2) and 63,870 (€ 52,785) (Figure 3, P1) in an attempt to be identical in terms of costs for
the farmer to use M1 or M2. Even though in these points M2 does not represent an increase
in farmer’s profit, there is an important reduction in herbicide usage, which signifies a
decrease in environmental impact.
Together with the growth of no–till in Argentina in the last years, it grew the chemical weed
control. In 1990 no-till was not a common practice and it was applied only to 62,000
hectares. However, this number rose up to 16,000,000 hectares in 2003. During this period,
the amount of herbicide traded has been increased in 710%. If we considered that no-till area
will continue to grow, the long term sustainability of the system should be questioned.
When using uniform spraying applications, there are important economic and environmental
consequences. Although these effects could be easily identified, both of them have a strong
connection. Herbicides are an important part of agricultural costs. Excessive or inappropriate
herbicide usage could generate modifications on agronomic ecosystem, causing a significant
change in weed population. What is more, this might tuned out into the appearance of weeds
which are tolerant or resistant to the dosage applied. About 235 tolerant or resistant weed
species spread in 42 countries have been discovered (International Survey of Herbicide-
Resistant Weeds 2000). On the other hand, this higher dosage necessary for controlling
these tolerant or resistant weeds generates a higher cost and an inefficient usage of the
input.

Figure 3: Maximum and minimum price for weed detection system. (US dollars)

Consequently, farmer production cost is increased. In addition to this, uniform spraying


spreads the product over land where there is no weed coverage at all, polluting soil and
underground water, just as a result of increasing the treated area. The main advantages of
this weed detector spraying system leis on these statements. Even though this technology
does not have the power of improving the spraying technique applied, it is possible by its
usage to decrease the treated area by this active ingredient. We should also consider that
this inappropriate herbicide use may pose health risks (Wiles et al. 1992).

Conclusion
Although the method would be extremely useful for argentine farmers in order to reduce
environmental impact and increase their profit, the imported weed detection system is only
convenient if there is low weed coverage and if a high quantity of hectares are worked yearly.
Due to argentine agro economic scenario, the price of the system should be between the
ranges established above and this might be achieved if the system is manufactured locally.
References
Brown R. B.; Andreson G. W.; Proud B. and J. P. Stekler. 1990. Herbicide application control using GIS weed
maps. St. Joseph, Mich., American Society of Agricultural Engineers Paper Nº 00-1121, 12 p.
Cardina J., Sparrow D. H. and E. L. McCoy. 1995. Analysis of spatial distribution of common lambs-quarters
(chenopodium album) in no-till soybean (Glycine max). Weed Science, 43, 258-268.
Gerhards R., Wyse-Pester D. Y., Mortensen D. and G. A. Johnson. 1997. Characterizing spatial stability of weed
populations using interpolated maps. Weed Science, 45, 108-119.
Johnson G. A.; Mortensen D A; and A. R. Martin.1995. A simulation of herbicide use based on weed spatial
distribution. Weed Research, 35, 197-205.
Moltoni, A. and L. Moltoni. 2005. Pulverización selectiva de herbicidas: implicancias tecnológicas y económicas
de su implementación en la Argentina. En VIII Congreso Argentino de Ingeniería Rural. Ed. CD-Rom, 9 al
12 de noviembre de 2005. Villa de Merlo, SL, Argentina.
Rew L J; Cussans G W; Mugglestone M A and P. C. H. Miller. 1996. A technique for mapping the spatial
distribution of Elymus repens with estimates of the potential reduction of herbicide usage from patch
spraying. Weed Research, 36, 283-292.
Shearer S. A. and R. G. Holmes. 1990. Plant identification using color co-occurrence matrices. Transactions of
the ASAE 34 (4), 1897-1909.
Thomson J P; Stafford J V and P. C. H. Miller. 1991. Potential for automatic weed detection and selective
herbicide application. Crop Protection, 10, 254-259
Tian L., Slaughter D. C. and R. F. Norris. 1997. Outdoor field machine vision identification of tomato seedlings for
automated weed control. Transactions of the ASAE 40 (6), 1761-1768.
Tian L., Reid J. and J. Hummel. 1999. Developement of a precision sprayer for site-specific weed management.
Transactions of the ASAE 42 (4), 893-900.
Wang N., Zhang F. E., Sun Y. and D. E. Peterson. 2001. Design of an optical weed sensor using plant spectral
characteristics. Transactions of the ASAE 44 (2), 409-419.
Wiles L. J., Wilkerson G. G., Gold H. G. and H. D. Coble. 1992. Modeling weed distribution for improved
postemergence control decisions. Weed science 40, 546-553.
Woebbecke D. M., Meyer G. E., Von Bargen K. and D. A. Mortensen. 1995. Shape feature for identifying young
weeds using image analysis. Transactions of the ASAE 38 (1) , 271-281.
Zhang N. and C. Chaisattapagon. 1995. Effective criteria for weed identification in wheat fields using machine
vision. Transactions of the ASAE 38 (3), 965-974
International Survey of Herbicide-Resistant Weeds. 2000. Herbicide Resistance Action Committee [on line].
Available at : < http://www.wssa.net>

You might also like