Journal of Archaeological Science 39 (2012) 2541e2551

Contents lists available at SciVerse ScienceDirect

Journal of Archaeological Science

journal homepage: http://www.elsevier.com/locate/jas

Mapping ancient chinampa landscapes in the Basin of Mexico: a remote sensing

and GIS approach
Christopher T. Morehart
Department of Anthropology, Georgia State University, Sparks Hall, Suite 335, 33 Gilmer St. Atlanta, GA 30303, USA

a r t i c l e i n f o a b s t r a c t

Article history: This paper uses remote sensing data to document a raised field, chinampa system adjacent to the
Received 3 January 2012 Postclassic kingdom of Xaltocan in the northern Basin of Mexico. Various forms of landscape informa-
Received in revised form tion; historic records and maps as well as remote sensing; are considered to understand the chinampa
28 February 2012
system. The remote sensing data examined include 1950s aerial photographs, Landsat 7 data, and
Accepted 2 March 2012
Quickbird VHR, multi-spectral imagery. This article evaluates the utility of each of these forms of data to
identify buried chinampa features and integrates them in a GIS to produce a map of Xaltocan’s chinampa
Keywords:
landscape. Canals of various sizes and hydrological positions comprised the chinampas and integrated
Chinampas
Remote sensing
the system together. Occupying at least 1500e2000 ha, Xaltocan’s chinampa system represents the
Satellite imagery largest pre-Aztec, chinampa system in the Basin of Mexico.
Landscape archaeology Ó 2012 Elsevier Ltd. All rights reserved.
Agriculture
Basin of Mexico
Xaltocan

1. Introduction Following Armillas, this paper uses remote sensing data to

Almost 40 years ago, Pedro Armillas (1971) published one of the
first reports on chinampa agriculture in the Basin of Mexico that
resulted from archaeological research. Theoretically, he attempted
to introduce an explicit landscape perspective into Mesoamerican
archaeology. Armillas envisioned an approach that was not simply
settlement or environmental archaeology but one wedded to the
materiality of the landscape itselfda palimpsest on which the
imprints of human action “are continually being erased and
rewritten, and quite often smudged” (Armillas, 1971:665).
Armillas used remote sensing data, aerial photos, to initiate his
research on Aztec chinampas in the southern Basin of Mexico:
“Methodologically, the archaeological investigation of a cultural
landscape begins with the interpretation of aerial photographs”
(Armillas, 1971:665). This approach influenced the methodologies
of central Mexican surveys of the 1960s and 1970s (e.g., Sanders
et al., 1979). Since then, archaeologists commonly use remote
sensing data in investigations on early hydraulic systems in central
Mexico (e.g., Frederick et al., 2005; Morehart, 2009; Nichols, 1988;
Nichols et al., 1991).
E-mail address: cmorehart@gsu.edu.
document a chinampa system adjacent to the Postclassic kingdom
of Xaltocan in the northern Basin of Mexico. Xaltocan was a political
center in the Basin of Mexico before the Aztec empire (Fig. 1). Like
Tenochtitlan, the later Aztec capital, Xaltocan was an island
kingdom. Brumfiel (2005) documented its initial settlement during
the 10th century AD, a time archaeologists refer to as the Early
Postclassic period. By the 14th century (Middle Postclassic period),
Xaltocan was an influential city-state. Like many other political
communities during the Middle Postclassic, Xaltocan was
embroiled in conflict. By the end of the 14th century, an alliance of
kingdoms conquered Xaltocan. The population fled. The town was
re-settled after the Aztec empire incorporated the northern basin
40 years later.
Unlike the relict chinampas Armillas observed, Xaltocan’s chi-
nampas are completely buried by a layer of eolian soil, with
virtually no visible topographic relief with the exception of some
locations where the largest canals created minor depressions
(Frederick et al., 2005; Morehart, 2009). In such situations, remote
sensing data are valuable tools for discovering past agricultural
systems (Lasaponara and Masini, 2011). This article evaluates the
usefulness of various forms of landscape imagery and integrates
them into a GIS to produce a map of Xaltocan’s chinampa system.
First, I discuss and define chinampa agriculture as a localized
manifestation of the broader technology of raised field agriculture.

0305-4403/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2012.03.001
2542 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551
Fig. 1. A regional map showing (1) the Basin of Mexico and (2) its location in Mexico as well as (3) Xaltocan and selected sites. The polygon displays the specific study area
(displayed in detail in Fig. 8).
Second, I provide an overview of available historical data on chi-
nampas and other hydraulic systems around Xaltocan. Much of this
historical information elucidates the anthropogenic landscape of
Xaltocan and provides a crucial starting point for subsequent
remote sensing (and field) investigations. Next, I examine the utility
of 1950s aerial photographs, Landsat data, and Quickbird VHR
imagery. These data were used to create a map of the chinampa
landscape, which provides an opportunity to consider the farming
system’s size and structure. I conclude with a consideration of some
of the broader social, political, and economic dimensions of this
work as well as the strengths and weaknesses of the remote
sensing data used.
2. Background
2.1. Chinampa agriculture
Chinampas are one of the most widely discussed yet least
archaeologically studied forms of prehispanic agriculture. The
number of archaeological projects that have examined chinampas
has grown (Avila López, 1991, 2006; Parsons et al., 1982, 1985;
Frederick et al., 2005; Morehart, 2009). However, most information
continues to come from indirect means: historic texts, quasi-mythic
narratives, ethnographic descriptions, and ecological research (e.g.,
Coe,1964; Crossley,1999; Gomez Pompa and Jiménez Osornio,1989;
Palerm, 1973; Rojas Rabiela, 1991; Sanders, 1957; Santamaría, 1912;
West and Armillas, 1950).
The term chinampa derives from the Nahuatl word chinamitl,
meaning an area enclosed by a hedge or canes (Molina, 1944:21).
Despite the specificity of this definition, the term is used by both
scholars and agriculturalists to designate agricultural land in
a wetland environment in which plots are elevated above water
levels and surrounded by canals. Chinampa fields are typically
narrow, around 4 m wide, but may extend in length up to
400e900 m (Santamaría, 1912:13). Often willows or cypress trees
are planted along the edges to protect the banks from erosion.
As a form of raised field agriculture chinampas share similarities
with comparable systems in highland and lowland regions of the
New World. Raised field have been documented elsewhere in
highland Mexico as well as in the lowlands of the Mexican Gulf
Coast, northern Belize, and Guatemala (e.g., Denevan, 1970, 1982;
Doolittle, 1990; Farrington, 1985; Fisher, 2005; Pohl, 1990; Pohl
et al., 1996; Puleston, 1978; Puleston and Siemens, 1972;
Scarborough, 2003; Siemens, 1983; Turner and Harrison, 1983;
Whitmore and Turner, 2001; Wilken, 1987). In South America,
raised fields characterize landscapes in the Andean highlands and
the Amazonian lowlands (e.g., Bandy, 2005; Darch, 1983; Denevan,
2001; Erickson, 1993, 1994, 2006; Janusek and Kolata, 2004; Kolata,
1991; Stanish, 1994, 2006; Walker, 2011; Zimmerer, 1991).
Although the physical characteristics of raised fields vary intra- and
 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551
inter-regionally, they share some similarities. Unlike other agri-
cultural landscapes, too much rather than too little water is the
major constraint for raised field systems, including chinampas
(Jiménez-Osornio, 1999). Raised fields are known for high produc-
tivity and sustainability even under intensive and continuous
cultivation. Soil from canals is used as fertilizer, and the local
aquatic environments are rich in resources that are harvested in
conjunction with farming.
Raised field systems occupy central positions in “top-down” and
“bottom-up” debates on the political economy of agriculture (e.g.,
Erickson, 1993, 2006; Janusek and Kolata, 2004; Morehart and
Eisenberg, 2010; Stanish, 1994; Walker, 2011). In fact, chinampas
provide a key historical example of pre-Columbian state influence
over agriculture. Prior to the formation of the Aztec empire in AD
1428, chinampas existed on a limited scale along the edges of lakes
Xochimilco and Chalco in the southern Basin of Mexico (Parsons
et al., 1982). After the empire formed, the Aztecs implemented
systematic chinampa agriculture in this region. The highly stan-
dardized chinampa layout over approximately 12,000 ha suggests
state administration in their planning and construction (Armillas,
1971:660). These chinampas supported up to one-half of the pop-
ulation Tenochtitlan, the Aztec capital (Armillas, 1971; Parsons,
1976).
2.2. Xaltocan’s chinampas in the historical record
Chinampa agriculture’s association to Early and Middle Post-
classic period political entities, such as Xaltocan, has received less
attention than later Aztec (Late Postclassic period) chinampas.
However, the existence of ancient chinampas at Xaltocan has been
known for several decades and even centuries. The early 17th
century chronicler Tezozómoc (1975:38) wrote that the Mexica
(who would become a powerful ethnic group in the Aztec Triple
Alliance) witnessed chinampas in Xaltocan at the end of the 12th
century A.D. during their mythic migration into central Mexico.
Fig. 2. Selected historic maps of Basin of Mexico depicting Xaltocan and the Ozumbilla springs. Maps from: a. 1555 (Santa Cruz); b. 1607; c. 1774; d. 1774. Ozumbilla source indicated
at arrows. Maps adapted from (Memoria de las Obras del Sistema Lacustre de Drenaje Profundo del Districto Federal, 1975, vol. IV).
2543
During the Spanish invasion, Cortés and Bernal Díaz del Castillo
described Xaltocan as a town with many channels and canals
(Palerm, 1973:37e39).
Discussion of chinampas is sporadic following the Spanish
conquest. The Descripción del Arzobispado de México, written in
1571, states that Xaltocan had farms adjacent to lagoons, which
might refer to chinampas (Paso y Troncoso, 1905a:35). Another
significant record is a document made by Luis Carrillo de Guzmán
in 1599, who visited the area to assess the capacity of local people
to pay tribute (Frederick Hicks, personal communication 2009).
According to this document, these areas were characterized by
“lands of chinampas and nitrate [salitrales] and other flat, plain
land in which they sow their fields, which are fertile and are fed
from a river that passes to the east of the town and from the sources
of water that come from Ozumbilla” (AGN Tierras 1584/1 f.4v-5r).
Ozumbilla, near an area also referred to as Ojo de Agua, would
have provided one of the few freshwater sources in the brackish
lacustrine environment (Frederick et al., 2005).
Carrillo de Guzmán also described chinampas in other nearby
locations, such as San Joan Atenango, San Pedro Miltengo (Mil-
tenco), Santa María Tenamitlan (Tonanitla), and Santa Ana Nes-
tlalpa (Nextlalpan). The Suma de Visitas, written in the mid-16th
century, records that Nextlalpan had two harvests a year for both
wheat and maize (one on irrigated land and the other on dry) (Paso
y Troncoso, 1905b:166). Due to regional disparities in land quality,
“chinampas from this town [Xaltocan] are divided up because those
that they have are flooded...and thus each town is given [land]
according to the number of people” (AGN Tierras 1584/1 ff. 30e31).
The reference to Colonial towns using Xaltocan lands for agriculture
may reflect the role of Xaltocan, then municipal capital (cabecera),
in administering land resources to local residents not technically
residing in the town itself.
Unlike the chinampas, the Ozumbilla springs are less ambig-
uous. They appear in several historic maps (Fig. 2). The Relación
Geográficas de Chiconautla states that “in this jurisdiction there are
2544 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551
five rich sources from which water flows” (Paso y Troncoso,
1905c:172). The source and canals that led from it are also
described in detail by Carillo de Guzmán. He not only describes
a canal that led to Xaltocan from the source but that another major
canal branched off and headed south toward Tonanitla:
They say that.this arroyo of water comes from the slopes of
a bare hill [Chiconautla] where there are some small trees.and
that the source from which the arroyo emerges is called the
sources of Ozumbilla, and that this hill is to the southeast.and
before arriving in the said town [Xaltocan] a branch of the said
arroyo splits off and goes to the town of Santa María [Tonanitla]
(AGN Tierras 1584/1, f. 6v).
The source also figures in the history of the drainage of the Basin
of Mexico’s lakes. Don Luis de Velasco, Viceroy of New Spain (AD
1550e1564), ordered the diversion of the springs at Ozumbilla to
create an artery to bring supplies to Mexico City and to alleviate the
bad odor of the drying lake (González Obregón et al., 1902:69).
Velasco’s efforts never got off the ground, but they represent the
initial planning efforts to drain the basin’s lakes: the Desagüe. The
springs and associated drainages were functioning in the 19th
century. Orozco y Berra (1864:169e170) discussed the source
principally in terms of fishing, suggesting a shift in the local
economy. The springs can also be seen in 1950s aerial photos (see
below), though they no longer exist.
2.3. Rediscovering Xaltocan’s chinampas
Frederick conducted the first archaeological study of Xaltocan’s
chinampa system in the late 1980s (Frederick et al., 2005). Using
aerial photographs made by the Compañía Aerofoto Mexicana in
1956, Frederick was able to partially map and estimate the scale of
the system. In the aerial photos, the chinampas are visible as canals
of various sizes and orientations (Fig. 3). Given the extent of the
chinampa system observable in the photos, he estimated the size of
the system to be between 100 and 200 ha.
Frederick identified the major canal that brought water into the
chinampa system from the Ozumbilla springs (Fig. 3). This feature is
clearly visible in the aerial photos and was between 50 and 80 m
Fig. 3. Aerial photograph of section of Xaltocan’s chinampa system. Primary canals at
arrows. Region that Frederick studied in box.
wide. It was critical in changing the hydrological nature of the lake
environment, creating a less saline environment (Frederick et al.,
2005:107).
3. Remote sensing data
The historical data and Frederick’s pioneering work constitute
key sources of data on Xaltocan’s ancient chinampa landscape.
However, the spatial extent and the structure of the chinampa
system were still not well known prior to this research. To do so, I
examined several different kinds of remote sensing data: (1) the
same aerial photos Frederick examined; (2) Landsat data; and (3)
Quickbird VHR satellite data. The utility of these data were
compared, and they were integrated into a Geographic Information
System (ArcGIS) to create a map of Xaltocan’s chinampas.
3.1. Aerial photos
The 1950s aerial photographs of the northern Basin of Mexico
were incorporated in ArcGIS to reassess the spatial extent of agri-
cultural features (Fig. 3). The images were scanned at 3200 dpi
using an Epson Photo Flatbed Scanner, which produced sufficiently
high resolution images to re-examine the landscape (Fowler and
Fowler, 2005). The images were georeferenced using GPS ground
control points collected in the field at historically persistent places
(i.e., ranches, haciendas, or churches). A 95% linear Gaussian
enhancement was applied to the imagery to accentuate the
contrast between pixels prior to incorporating them into ArcGIS.
Based on this work, Frederick’s estimate of the size of the chi-
nampa zone was too conservative: the area seemed to be at least
400e500 ha. Given that the springs were at least about 2.5 km
south of the chinampas he documented, it is likely that the chi-
nampas occupied an even larger area (see Fig. 8, lower right inset).
The system was composed of canals of various sizes: the principal
canal Frederick identified (Fig. 3, white arrows), smaller canals
leading off from it, and narrow canals probably separating indi-
vidual fields. Another major canal splits off from the principal canal
south of Xaltocan and heads south (Fig. 3, black arrows). This canal
is most likely the same one Carrillo de Guzmán described that
branched off and connected Xaltocan to Tonanitla. Moreover, a long
narrow canal can be seen heading south from Tonanitla, though it
does not appear to be connected to any other branching canals
(Fig. 8, lower left inset).
3.2. Landsat data
Landsat 7 ETM þ data were acquired and incorporated into the GIS
to assess the applicability of satellite imagery. This imagery is
comprised of 7 bands from the visible to the mid-infrared parts of the
spectrum. Each band has an Instantaneous Field Of View (IFOV) of
30 m, with the exception of band 6 (thermal infrared), which has an
IFOV of 60 m. These data also include a panchromatic image with an
IFOV of 15 m. Thus, particular band combinations can be pan-
sharpened to produce multi-spectral images with a resolution of
15 m per pixel (Parcack, 2009). Pan-sharpened composites were
produced using ArcGIS’s raster processing tools. Fig. 4 presents a False
Color, pan-sharpened composite made by assigning bands 4, 3, and 2
to the red, green, and blue color channels of ArcGIS (4-3-2 RGB).
The Landsat data show the major branching canal described in
historic sources, though the source canal from Ozumbilla is less
obvious. Additional possible canal features were detected to the
northwest of the chinampa zone. Regardless of the band combi-
nation employed, the resolution of the pan-sharpened data was not
high enough to detect smaller linear features. Nevertheless, this
endeavor demonstrated the potential of satellite data.
 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551
Fig. 4. False color composite of Landsat 7 ETM þ images (4, 3, 2-RGB).
3.3. Quickbird data
Archaeologists interested in identifying archaeological
features are increasingly employing very high resolution (VHR)
commercial satellite data (e.g., De Laet et al., 2007; Due Trier
et al., 2009; Garrison et al., 2008; Lasaponara and Masini, 2007,
2011; Parcack, 2009). To better document the chinampa system,
Quickbird VHR satellite data was obtained from Digital Globe
(acquired March 3, 2005, during the local dry season). The
imagery is multi-spectral, combining bands from the visible blue
(Band 1), visible green (Band 2), visible red (Band 3), and near
infrared (Band 4) parts of the spectrum. Like Landsat 7, Quickbird
data also come with a panchromatic band. The multi-spectral
images each have an IFOV of 2.5 m, whereas the panchromatic
image is 0.60 m. Specific band combinations can be pan-
sharpened to produce composite images with a resolution of
60 cm. Digital Globe delivered the imagery with a standard
georeference, which was checked using ground control points
and found to be adequate given the flatness of the terrain and the
relatively restricted spatial scale of the study area.
Fig. 5 displays all four multi-spectral images individually over
a section of the coverage. Canal features are clearly visible. Different
bands accentuate specific surface attributes to varying degrees. For
example, the very light areas in the near infrared image, especially
the rectangular parcels, indicate the presence of green vegetation
(or organic-rich irrigation water) because vegetation reflects
strongly in the near infrared part of the spectrum (Lillesand and
Kiefer, 1994). This band can be a useful if subsurface features
affect the vitality of vegetation (Lasaponara and Masini, 2007;
Parcack, 2009:91). In examining each individual band, however, it
appears that none captures chinampa features better than any
other.
Fig. 6a and b shows a pan-sharpened false color composite and
a pan-sharpened true color composite of an area in the image.
Despite differences in how each composite accentuates surface
characteristics, little difference can be seen. In fact, isolated chi-
nampa features were almost entirely overwhelmed by strong
2545
reflectance in the near infrared in many areas, especially in irri-
gated farm parcels. A normalized difference vegetation index
(NDVI) can be used to better assess how buried features affect
vegetation vigor (Lasaponara and Masini, 2007). I calculated a NDVI
with the multi-spectral data (visible red and near infrared bands)
using ArcGIS’s raster calculator (Fig. 6d). This technique empha-
sized areas accentuated by the near infrared band but almost
completely obscured canal features. The low utility of the NDVI and
the NIR alone, suggests that the NIR band is not useful for high-
lighting buried chinampa features. Several band combinations
were tried but none had a global impact on accentuating archae-
ological features. In fact, the panchromatic image alone can be
usefully employed to study the extent of chinampa features
(Fig. 6c).
3.4. Mapping chinampas
The primary objective was to create a map to understand the
scale and structure of the farming system. Because Quickbird
imagery has a high resolution, it is an ideal source for mapping the
agricultural landscape. Given the high degree of “noise” and the
lack of a distinctive spectral signature for buried agricultural
elements, digital feature extraction and edge enhancement tech-
niques were not very successful except at extremely restricted
spatial scales. Even combining a low pass filter to reduce noise and
a high pass filter to enhance feature edges (Lasaponara and Masini,
2011:196) was productive only in localized areas. In order to map
the chinampa landscape, however, it was often necessary to
undertake such highly localized techniques to accentuate features.
Envi’s Region of Interest tool (ROI) was useful for such spatially
restricted tasks. Nevertheless, positive results of these analyses
were always re-incorporated into the ArcGIS map.
Potential features in the satellite images were compared to areas
in the aerial photos to better assess their identification as ancient,
providing a useful historical check. Some had to be excluded from
the map. For example, Fig. 7 shows a situation where a road that led
west from a ranch in the 1950s left a linear shadow after it was
shortened to meet the local highway (left) as well as a 20th century
field boundary that no longer exists but left a linear feature in the
landscape (right). This method reveals the continued utility of
historic aerial photos even in areas where VHR satellite images are
available.
4. The map
The map was made in ArcGIS by digitally drawing polyline
and polygon features over canals, using polygons for major canals
and lines for small canals adjacent to fields. Where identified,
parcels were mapped with polygon features. Fig. 8 shows a map
of the chinampa system. Its extent is much larger than Frederick
et al.’s (2005) or my estimate using aerial photos: 1500e1600 ha
is an acceptable (and likely conservative) estimate. It may be
larger and extend farther south into the farm land of Tonanitla.
An estimate of 2000e2500 ha is probably more accurate. In fact,
if the area is expanded beyond the system’s primary canal
network to include the major canal that leads south from
Tonanitla and the springs (rather than just the canal emanating
from it), the system would approach 3500 ha in size. However,
I maintain the more conservative estimate between 1500 and
1600 ha.
The system’s structure can be characterized in terms of the size,
shape, and hydrological position of canals. There were three types
of canals, which I label primary, secondary, and tertiary. Two
primary canals existed. One led from the springs at the Ozumbilla
source. The other, discussed above as the branch, departed this
2546 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551

Fig. 5. Four, individual multi-spectral band images comprising the Quickbird multi-spectral imagery.

Fig. 6. Comparison between (a) true color composite (3, 2, 1-RGB); (b) a false color composite (4, 3, 2-RGB); (c) the 60 cm panchromatic image; and (d) the results of a normalized
difference vegetation index. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551 2547

Fig. 7. Comparison of Quickbird imagery (top) and aerial photos (bottom) showing remnant of features from the mid-20th century landscape. The left images show a clear road
heading east from a ranch and crossing a local highway in the aerial image (bottom) whose remnant can be seen in the satellite image (top). The right images show a road dividing
agricultural parcels in the 1950s (bottom) whose imprint can still be seen in the satellite images (top). The arrows indicate the features.

canal south of Xaltocan and connected with Tonanitla. The canal
connected to the Ozumbilla springs was approximately 4 km long,
whereas the canal between Xaltocan and Tonanitla was almost
3 km in length. These canals were between 40 and 60 m wide.
The major functions of primary canals were as central fresh-
water arteries. The primary canal leading from the Ozumbilla
springs was more hydrologically significant, bringing fresh,
spring water into the brackish environment (Frederick et al.,
2005:93). Given the degree of land modification in this region,
it is not possible to estimate the amount of water that flowed
from these springs. The other primary canal, the “branch,” is
classified as a primary canal due to its size and linear nature. Its
hydrological role in the chinampa system was secondary to the
Ozumbilla canal. It likely was a transportation artery between
Xaltocan and Tonanitla to the south, one of Xaltocan’s possible
tributaries during the Middle Postclassic. Lastly, the linear canal
heading south from Tonanitla may be another transportation
artery connecting this area to communities farther south, such as
Ecatepec.
The principal role of secondary canals was to distribute water
from the primary canals into the heart of the chinampa system. In
the central area, where they branched off from the primary canals,
they formed a complicated network, unlike the standardized chi-
nampa system that the Aztec’s would later sponsor in the south.
Farther to the east, however, secondary canals appear more orga-
nized, possibly reflecting a need for greater network simplicity
given their distance from the primary canal. Although these canals
likely constitute a hydrological network in a topological sense, it is
problematic to undertake a quantitative analysis of network
complexity and connectivity with incomplete data (Haggett and
Chorley, 1969).
Tertiary canals were “feeder” canals that separated individual
fields. These are the long, narrow canals that give chinampas their
characteristic appearance. They operated at the most localized
spatial scale, providing water and access to individual fields. Most
tertiary canals led off from secondary canals, though some were
connected to primary canals. It was possible to map 1017 tertiary
canals of various sizes. In areas where they are visible in the
remote sensing data, the canals were between 1.5 and 2 m wide
and the adjacent field surfaces were about 4 m, which is sup-
ported by excavation data on the chinampas (Morehart, 2009). On
average, the mapped canals were 49.05 m long, though they have
a standard deviation of 25.54 m and a range of 127 m. In some
instances, canals had short lengths because newer canals trun-
cated them. However, mapping tertiary canals is highly affected by
preservation and visibility. They are the narrowest and shallowest.
In areas that have been plowed, they are not visible and cannot be
mapped.
Xaltocan’s chinampa system appears to have been parceled
into groups of fields. Parcels occupied a level of land integration
above individual fields. They are delineated either by the pres-
ence of secondary canals that divide one parcel from another or
by an abrupt reorientation of tertiary canals. The upper right inset
of Fig. 8 shows a unique case where one large parcel (1.25 ha) had
groups of fields with clearly distinct orientations bundled by
a distinct secondary canal. Fig. 9 shows the chinampa system
highlighting the distribution of mapped parceled lands. It was
possible to document 422 parcels. The number of fields in each
parcel visible varies, ranging from three to four to over twenty.
The lengths of fields in each parcel differ also, but generally larger
parcels had longer fields. Sometimes it was possible to map
parcels but not internal fields, and in other instances tertiary
canals but not parcels. Notwithstanding the unavoidable gaps, the
high degree of variability in parcel size is notable. The average
area of a parcel is 0.25 ha, but the standard deviation is large,
0.30, and the range is 2.63. Nevertheless, caution must be
employed when interpreting these figures given limitations in
data coverage.
2548 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551

Fig. 8. Map of Xaltocan’s chinampa system showing canal typology. The upper right inset is a close-up of Quickbird’s panchromatic image. The lower left inset shows the canal
south of Tonanitla in the aerial photos. The lower right inset shows the springs in the aerial photos, which still existed in the 1950s. This inset, however, is not offset but is in its
actual location in reference to the other mapped landscape features.

5. Discussion and conclusions
This report is focused only on exploring the utility of remote
sensing data to identify and map the chinampas of Xaltocan,
Mexico. But it is part of a larger project on the sociopolitical
dynamics of chinampa agriculture (Morehart, 2009, 2010). Multiple
excavations in the chinampa zone yielded unequivocal evidence of
buried agricultural features. Although historical records suggest the
continued use of these features following the Spanish conquest,
this claim is not certain. Virtually all chronological data (AMS dates
and pottery) date to the Middle Postclassic (AD 1200e1400), when
Xaltocan was an independent polity. The chinampa landscape
seems to have been mostly abandoned after the kingdom’s
conquest. Later drained fields probably existed but on a more
limited scale. The major canals no doubt continued in use, however,
which may explain the historical records.
The integrated system of fields and canals described above likely
represents the largest pre-Aztec chinampa system in central
Mexico. This study provides dramatically larger estimations of the
size and integration of Xaltocan’s chinampa system than previous
approximations, results with significant implications for under-
standing the sociopolitical implications of agriculture. The global
structure of Xaltocan’s chinampas is as much an integrated, irri-
gation system as it is a lacustrine, raised field landscape (cf. Palerm,
1973). In a lake setting with abundant freshwater, water scarcity
may not create organizational problems requiring a high degree of
coordination (Erickson, 1993). Where freshwater is scarce or is
confined to specific sources, such as at Xaltocan, farmers must
coordinate their activities at greater levels of integration, which
may involve the co-participation not only of individual households
but of entire communities (Morehart, 2010:86). Together with
integration, the size of the system also would call for some degree
of supra-household administration (Hunt, 1988).
The combination of (1) large size and (2) a high degree of
integration can make such socio-ecological systems susceptible to
influence by pre-existing or emergent power-brokers not simply
due to their physical characteristics but due to their social ones
(Morehart, 2010:90e91). “Cooperative relationships are one way in
 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551
Fig. 9. Map of chinampa zone showing the distribution and sizes of mapped parcels.
which local hierarchy is constructed and power exercised” (Cohen,
1999:7). The chronological correlation between Xaltocan’s chi-
nampas and its existence as an independent polity suggests
a degree of dependence between the agricultural landscape and the
political community. This aspect of Xaltocan’s chinampas contrasts
with the later Late Postclassic chinampas in the southern Basin of
Mexico. These chinampas were not highly integrated but were
influenced by the Aztec state (Sanders and Price, 1968:177), sug-
gesting differences between these two cases in the mechanisms of
integration of political institutions with the local institutions that
govern farming landscapes.
The previous discussion of the chinampa landscape clearly
displays a territorial bias, yet this was an aquatic agricultural
system. Water was critical to the hydrological functioning and cycle
of the system. The presence of canals allowed plants to be watered
systematically via pot irrigation, using a pole with a bag attached
(called a zoquimaitl in Nahuatl), or by splashing water onto plants
with canoe paddles (Sanders, 1957:78e80; Santamaría, 1912:Fig. 3;
West and Armillas, 1950:176; Wilken, 1987:170). Many scholars
have argued that the proximity of field surfaces to canals allows the
natural infiltration of water toward the center of the plots through
the capillary fringe of upper soil layers, referred to as sub-irrigation
(Rojas Rabiela, 1995:54; Sanders, 1957:76). Crossley (1999),
however, observed that rather than facilitating sub-irrigation,
water levels create warmer and more stable micro-climatic
temperatures that protect crops from frosts that are common in
this region during the dry season (see also Kolata and Ortloff, 1989).
Chinampa farming is closely integrated into local lacustrine
environments. Lacustrine resources in central Mexico were
important elements in both subsistence and market economies
(Parsons, 2006; Sahagún, 1963). Up to the 20th century, residents
collected fish and fish eggs, birds, insects, and other animals. Arti-
facts recovered from the chinampas, such as ceramic net weights
for fishing or bird-hunting and ceramic pellets for hunting, suggest
that past farmers undertook some of these activities (Morehart,
2010:268e273). The lakes were also sources for important non-
domesticated flora. For example, between the 16th and 18th
2549
centuries, residents of Xaltocan and neighboring towns supple-
mented their incomes by making reed mats (Gibson, 1964:336).
They likely integrated these activities with their agricultural strat-
egies to obtain items for daily subsistence, for feasts or ritual
events, or as goods to be traded in the market.
Different sources of remote sensing imagery were evaluated for
this project. Landsat data are limited by their resolution. Aerial
photographs and Quickbird multi-spectral and panchromatic
images, however, are effective, especially given historical changes
over the past 50 years. However, Quickbird data also had limita-
tions. Grøn et al. (2011:2030) recently observed that “it is important
to be aware...that spatial variations occurring within similar types
of features due to ploughing, individual behavioral patterns in
prehistory as well as variations due to the geological background
noise can cause considerable variation in the way one type of
archaeological feature will appear.” A mosaic of land-use practices
surrounds Xaltocan. This diversity creates a considerable degree of
“noise” that limits the global application of particular image anal-
yses over the whole area. Mapping the system requires a technique
that is flexible but is ultimately reliant on the knowledge of the
individual analyst.
Specific band combinations were not very useful in dis-
tinguishing ancient agricultural features, including NDVI. This may
be due to (1) the fact that the imagery was acquired in the dry
season when most fields were bare and when non-agricultural
vegetation was dormant, (2) the likelihood that buried features
have little effect on vegetation, (3) the possibility that the signa-
tures for the buried features have wavelengths located in parts of
the spectrum not available in Quickbird data, or (4) that features
have more “micro-level” spectral signatures that can only be dis-
cerned with hyperspectral data. Other archaeologists have had
success in using specific band combinations and a NDVI to enhance
past features (e.g., Lasaponara and Masini, 2007). Thus, imagery
acquired in the rainy season may reveal associations between
vegetation vigor and buried chinampas. However, many of these
studies occurred (1) at more localized scales (2) on a small number
of features (3) in a restricted area. Employing remotely sensed
2550 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551
topographic data is another option to map the landscape. High
resolution LIDAR data, for example, represents a costly alternative
that is not currently available for the study area. Micro-topographic
field mapping may be useful too but likely only in small areas given
the greater time investment compared to remote sensing. The
application of both these methods is planned as a part of continuing
work on the chinampas and the historical ecology of the northern
Basin of Mexico.
Despite some limitations, remote sensing imagery, particularly
VHR satellite data, offered an effective strategy to identify Xaltocan’s
chinampa system, to map its extent, and to document its hydrological
structure. This work provides a rare glimpse of an ancient farmscape
that not only was an economic and ecological backbone of a broader
political community. It was also a space where many past people spent
much of their daily lives.
Acknowledgements
This research was supported by a National Science Foundation
Dissertation Improvement Grant, a Wenner Gren Dissertation
Grant, a Fulbright Hays Dissertation Research Fellowship, as well as
by several grants from Northwestern University. Fieldwork carried
out by the Proyecto Chinampero Xaltocan was made possible by
a permit by Mexico’s Instituto Nacional de Antropologia e Historia.
Charles Frederick kindly provided access to the aerial photos he
used. I would like to thank Elizabeth Brumfiel for her support and
encouragement throughout the work.
References
AGN Tierras 1584/1. Testimonio de los autos de reducción que los naturales de los
pueblos sujetos se hizo al de Xaltocan su cabezera, juron de Sumpango de la
Laguna, el año de 1599. Ramo de Tierras, vol. 1584, No. 1. Archivo General de la
Nación, Mexico, D.F.
Armillas, P., 1971. Gardens on swamps. Science 175, 653e661.
Avila López, R., 1991. Chinampas de Iztapalapa, D. F. Instituto Nacional de Antro-
pología e Historia, Mexico, D.F.
Avila López, R., 2006. Mexicaltzingo: Arqueologia de un Reino Culhua-Mexica.
Conaculta, Mexico.
Bandy, M.S., 2005. Energetic efficiency and political expediency in Titicaca Basin
raised field agriculture. J. Anth. Arch. 24, 271e296.
Brumfiel, E. (Ed.), 2005. Production and Power at Postclassic Xaltocan. Instituto
Nacional de Antropología e Historia, Mexico, D.F.
Coe, M., 1964. The chinampas of Mexico. Sci. Am. 210e211, 90e98.
Cohen, J.H., 1999. Cooperation and Community: Economy and Society in Oaxaca.
University of Texas Press, Austin.
Crossley, P., 1999. Sub-Irrigation and Temperature Amelioration in Chinampa
Agriculture. PhD Dissertation, University of Texas, Austin.
Darch, J.P. (Ed.), 1983. Drained Field Agriculture in Central and South America. B.A.R.
International Series, 189. Oxford.
De Laet, V., Paulissen, E., Waelkens, M., 2007. Methods for the extraction of
archaeological features from very high-resolution Ikonos-2 remote sensing
imagery, Hisar (Southwest Turkey). J. Archaeol. Sci. 34, 830e841.
Denevan, W.M., 1970. Aboriginal drained field cultivation in the Americas. Science
169, 647e654.
Denevan, W.M., 1982. Hydraulic agriculture in the American tropics: forms,
measures, and recent research. In: Flannery, K. (Ed.), Maya Subsistence: Studies
in Memory of Dennis E. Puleston. Academic Press, New York, pp. 181e203.
Denevan, W.M., 2001. Landscapes of Native Amazonia and the Andes. Oxford
University Press, Oxford.
Doolittle, W.E., 1990. Canal Irrigation in Prehistoric Mexico: The Sequence of
Technological Change. University of Texas Press, Austin.
Due Trier, Ø., Larsen, S., Solberg, R., 2009. Automatic detection of circular structures in
high-resolution satellite images of agricultural land. Archaeol. Prosp. 16, 1e15.
Erickson, C., 1993. The social organization of prehispanic raised field agriculture in
the Lake Titicaca Basin. In: Scarborough, V., Isaac, B. (Eds.), Economic Aspects of
Water Management in the Prehispanic New World, Greenwich, CT,
pp. 369e426.
Erickson, C., 1994. Methodological considerations in the study of ancient Andean
field systems. In: Miller, N., Gleason, K. (Eds.), The Archaeology of Garden and
Field. University of Pennsylvania Press, Philadelphia, pp. 111e152.
Erickson, C., 2006. Intensification, political economy, and the farming community.
In: Marcus, J., Stanish, C. (Eds.), Defense of a Bottom-up Perspective of the Past.
Agricultural Strategies. Cotsen Institute of Archaeology, Los Angeles,
pp. 233e265.
Farrington, I. (Ed.), 1985. Prehistoric Intensive Agriculture in the Tropics. BAR
International Series, 232. Oxford.
Fisher, C., 2005. Abandoning the garden: demographic and landscape change in the
Lake Pátzcuaro basin, Mexico. Am. Anth. 107, 87e95.
Fowler, M., Fowler, Y., 2005. Detection of archaeological crop marks on declassified
CORONA KH-4B intelligence satellite photography of southern England.
Archaeol. Prosp. 12, 257e264.
Frederick, C., Winsborough, B., Popper, V., 2005. Geoarchaeological investigations in
the northern Basin of Mexico. In: Brumfiel, E. (Ed.), Production and Power at
Postclassic Xaltocan. Instituto Nacional de Antropología e Historia, Mexico, D.F,
pp. 71e116.
Garrison, T., Houston, S., Golden, C., Inomata, T., Nelson, Z., Munson, J., 2008.
Evaluating the use of Ikonos satellite imagery in Lowland Maya settlement
archaeology. J. Archaeol. Sci. 35, 2770e2777.
Gibson, C., 1964. The Aztecs under Spanish Rule: A History of the Valley of Mexico,
1519e1810. Stanford, University Press, Stanford.
Gomez Pompa, A., Jiménez Osornio, J., 1989. Some reflections on intensive tradition
agriculture. In: Truman, K., Gladwin, C. (Eds.), Food and Farm: Current Debates
and Policies. University Press of America, New York, pp. 231e253.
González Obregón, L., Espinosa, L., Díaz Lombardo, I., Esparza, R., 1902. Memoria
Histórica, Técnica, y Administrativa de las Obras del Desagüe del Valle de
México (1449e1900). Junta Directiva del Desagüe del Valle de México.
Grøn, O., Palmer, S., Stylegar, F., Esbensen, K., Kucheryavski, S., Aase, S., 2011.
Interpretation of archaeological small-scale features in spectral images.
J. Archaeol. Sci. 38, 2024e2030.
Haggett, P., Chorley, R., 1969. Network Analysis in Geography. St. Martin’s Press,
New York.
Hunt, R.C., 1988. Size and the structure of authority in canal irrigation systems.
J. Anth. Res. 44, 335e355.
Janusek, J.W., Kolata, K., 2004. Top-down or bottom-up: rural settlement and
raised field agriculture in the Lake Titicaca Basin, Bolivia. J. Anth. Arch. 23,
404e430.
Jiménez Osornio, J.J., 1999. Componentes Ecologicos Esenciales de la Tecnologia
Chinampera. In: Jiménez Osornio, J.J., Rorive, V.M. (Eds.), Los Camellones y
Chinampas Tropicales: Memorias del Simposio-Taller Internacional sobre
Camellones y Chinampas Tropicales, Ediciones, de la Universidad Autonoma de
Yucatan, Merida, pp. 120e130.
Kolata, A.L., 1991. The technology and organization of agricultural production in the
Tiwanaku state. Lat. Am. Ant. 2, 99e125.
Kolata, A.L., Ortloff, C., 1989. Thermal analysis of Tiwanaku raised field systems in
the late Titicaca basin of Bolivia. J. Archaeol. Sci. 16, 233e263.
Lasaponara, R., Masini, N., 2007. Detection of archaeological crop marks by using
satellite Quickbird multispectral imagery. J. Archaeol. Sci. 34, 214e221.
Lasaponara, R., Masini, N., 2011. Satellite remote sensing in archaeology: past,
present, and future perspectives. J. Archaeol. Sci. 38, 1995e2002.
Lillesand, T., Kiefer, R., 1994. Remote Sensing and Photo Interpretation, third ed.
Wiley, New York.
Molina, A., 1944. Vocabularo en Lengua Castellan y Mexicana. Ediciones Cultura
Hispanica IV. Colección de Incunables Americanos, Madrid.
Morehart, C. (Ed.), 2009. Proyecto Chinampero Xaltocan: Informe De La Temporada
De Campo 2007e2008. Report on File. Instituto Nacional de Antropología e
Historia, Mexico, D.F.
Morehart, C., 2010. The Archaeology of Farmscapes: Production, Place, and the
Materiality of Landscape at Xaltocan, Mexico. PhD Dissertation, Department of
Anthropology, Northwestern University, Evanston, IL.
Morehart, C., Eisenberg, D., 2010. Prosperity, power, and change: modeling maize at
Postclassic Xaltocan, Mexico. J. Anth. Arch. 29, 94e112.
Nichols, D., 1988. Infrared aerial photography and prehispanic irrigation at Teoti-
huacan: the Tlajinga canals. J. F. Arch. 15, 17e27.
Nichols, D., Spence, M., Borlanda, M., 1991. Watering the fields of Teotihuacán: early
irrigation at the ancient city. Anc. Mes. 2, 119e129.
Orozco y Berra, M., 1864. Memoria para la Carta Hidrografica del Valle de México.
Sociedad Mexicana de Geografía y Estadística, Mexico.
Palerm, Ángel, 1973. Obras Hidráulicas Prehispánicas en el Sistema Lacustre del
Valle de México. Instituto Nacional de Antropología e Historia, Mexico, D.F.
Parcack, S., 2009. Satellite Remote Sensing for Archaeology. Routledge, London.
Parsons, J., 1976. The role of chinampa agriculture in the food supply of Aztec
Tenochtitlan. In: Cleland, C. (Ed.), Cultural Change and Continuity: Essays in
Honor of James B. Griffin. Academic Press, New York, pp. 233e257.
Parsons, J., 2006. The Last Pescadores of Chimalhuacán, Mexico. Museum of
Anthropology, Anthropological Papers 96. University of Michigan, Ann Arbor.
Parsons, J., Brumfiel, E., Wilson, D., 1982. Prehispanic Settlement Patterns in the
Southern Valley of Mexico: The Chalco-Xochimilco Region. Memoirs of the
Museum of Anthropology 14. Museum of Anthropology University of Michigan,
Ann Arbor.
Parsons, J., Parsons, M., Popper, V., Taft, M., 1985. Chinampa agriculture and Aztec
urbanization in the Valley of Mexico. In: Farrington, I. (Ed.), Prehistoric Inten-
sive Agriculture in the Tropics. BAR International Series. Oxford.
Paso y Troncoso, F., 1905a. Descripción del Arzobispado de México. In: Papeles de
Nueva España. Segunda serie, Geografía y Estadística Tomo III. Impresores de la
Real Casa, Archivo de Indias en Sevilla, Museo Nacional de Méxco, Madrid.
Paso y Troncoso, F., 1905b. Suma de Visitas de Pueblos por orden alfabético. In:
Papeles De Nueva España. Segunda serie, Geografía y Estadística Tomo I.
Impresores de la Real Casa, Biblioteca Nacional de Madrid, Museo Nacional de
Méxco, Madrid.
 C.T. Morehart / Journal of Archaeological Science 39 (2012) 2541e2551
Paso y Troncoso, F., 1905c. Relaciones Geográficas de la Diócesis de México.
Impresores de la Real Casa, Real Academia de la Historia de Madrid y del
Archivo de Indias en Sevilla, Museo Nacional de México, Madrid, Madrid.
Pohl, M. (Ed.), 1990. Ancient Maya Wetland Agriculture: Excavations on Albion
Island, Northern Belize. Westview Press, Boulder.
Pohl, M., Pope, K., Jones, J., Jacob, J., Piperno, D., deFrance, S., Lentz, D., Gifford, J.,
Danforth, M., Josserand, K., 1996. Early agriculture in the Maya lowlands. Lat.
Am. Ant. 7, 355e372.
Puleston, D., 1978. Terracing, raised fields, and tree cropping in the Maya lowland:
a new perspective on the geography of power. In: Harrison, P., Turner, B.L.
(Eds.), Pre-Hispanic Maya Agriculture. University of New Mexico Press, Albu-
querque, pp. 225e246.
Puleston, D., Siemens, A., 1972. Ridged fields and associated features in southern
Campeche; new perspectives on the lowland Maya. Am. Antiq. 37, 228e239.
Rojas Rabiela, T., 1991. Ecological and agricultural changes in the chinampas of
Xochimilco-Chalco. In: Harvey, H. (Ed.), Land and Politics in the Valley of
Mexico. University of New Mexico Press, Albuquerque, pp. 275e290.
Rojas Rabiela, T., 1995. Las chinampas del Valle de Mexico. In: Rojas Rabiela, T. (Ed.),
Presente, Pasado y Futuro de las Chinampas. Centro de Investigaciones y
Estudios. Superiores en Antropologia Social, Hidalgo y Matamoros, Tlalpan,
pp. 53e70.
Sahagún, B., 1963. Florentine Codex, General History of the Things of New Spain: Book
11, Earthly Things. Monographs of the School of American Research, Sante Fe, NM.
Sanders, W., 1957. Tierra y Agua (Soil and Water): A Study of the Ecological Factors
in the Development of Meso-American Civilizations. PhD Dissertation, Harvard
University, Cambridge.
Sanders, W., Parsons, J., Santley, R., 1979. Basin of Mexico: Ecological Processes in
the Evolution of a Civilization. Academic Press, New York.
2551
Sanders, W., Price, B., 1968. Mesoamerica: The Evolution of a Civilization. Random
House, New York.
Santamaría, M., 1912. Las Chinampas del Districto Dederal. La Secretaria de
Fomento, Mexico.
Scarborough, V., 2003. The Flow of Power: Ancient Water Systems and Landscapes.
School American Research Press, Santa Fe.
Siemens, A., 1983. Wetland agriculture in pre-hispanic Mesoamerica. Geographical
Rev. 73, 166e181.
Stanish, C., 1994. The hydraulic hypothesis revisited: Lake Titicaca Basin raised fields
in theoretical perspective. Lat. Am. Ant. 5, 312e332.
Stanish, C., 2006. Prehispanic agricultural strategies of intensification in the Titicaca
Basin of Peru and Bolivia. In: Marcus, J., Stanish, C. (Eds.), Agricultural Strategies.
Costen Institute of Archaeology, Los Angeles, pp. 364e399.
Tezozómoc, F., 1975. Cronica Mexicayotl. Universidad Nacional Autónoma de
México, Mexico, D.F.
Turner, B.L., Harrison, P. (Eds.), 1983. Pulltrouser Swamp: Ancient Maya Habitat,
Agriculture, and Settlement in Northern Belize. University of Texas Press, Austin.
Walker, J., 2011. Social implications from agricultural taskscapes in the south-
western Amazon. Lat. Am. Ant. 22, 275e295.
West, R., Armillas, P., 1950. Las Chinampas de México, poesía y realidad de los
jardines flotantes. Cuad. Am. 50, 165e182.
Whitmore, T., Turner, B.L., 2001. Cultivated Landscapes of Middle America on the
Eve of Conquest. Oxford University Press, Oxford.
Wilken, G., 1987. Good Farmers: Traditional Agricultural Resource Management in
Mexico and Central America. University of California Press, Berkeley.
Zimmerer, K.S., 1991. Wetland production and smallholder persistence: agricul-
tural change in a highland Peruvian region. Ann. Assoc. Am. Geograph. 81,
443e463.