Jump to main content.

A Landscape Approach for Detecting and Evaluating Change in a Semi-arid Environment, San Pedro Watershed (U.S./Mexico)

Principal Investigator: William G. Kepner1

  1. U.S. Environmental Protection Agency, Office of Research and Development, Las Vegas, Nevada 89193-3478.

You will need Adobe Acrobat Reader, available as a free download, to view some of the files on this page. See EPA's PDF page to learn more about PDF, and for a link to the free Acrobat Reader.


Vegetation change in the American West has been a subject of concern throughout the twentieth century. Although many of the changes have been recorded qualitatively through the use of comparative photography and historical reports, little quantitative information has been available on the regional or watershed scale. It is currently possible to measure change over large areas and determine trends in ecological and hydrological condition using advanced space-based technologies. Specifically, this is being tested in a community-based watershed in southeast Arizona and northeast Sonora, Mexico using a system of landscape pattern measurements derived from satellite remote sensing, spatial statistics, process modeling, and geographic information systems technology. These technologies provide the basis for developing landscape composition and pattern indicators as sensitive measures of large-scale environmental change and thus, may provide an effective and economical method for evaluating watershed condition related to disturbance from human and natural stresses.

Landscape composition and pattern (distribution) affects key ecological transfer processes which govern the movement or flow of energy, nutrients, water, and biota over time and space. Land managers in the Southwest have traditionally been interested in status and trend in environmental conditions. The principal degradation processes that have occurred throughout the western rangelands involves 1) changes of vegetative cover which result in the introduction of exotic annual species or woody shrubs and trees, and 2) acceleration of water and wind erosion processes which result in soil loss and decrease water infiltration and storage potential. Historically, these have been linked to livestock grazing and short-term drought. However, rapid urbanization in the arid and semi-arid southwest within the last 25 years has become an important anthropogenic factor in altering land cover composition and pattern. Research relating to the quantification of landscape change over time will improve our ability to assess current and past condition and provide information and tools which can be employed for future watershed management throughout the West.


We believe that landscape composition and pattern measures are diagnostic of ecological and hydrological condition and thus can be utilized as an effective method for characterizing landscape vulnerability to disturbance associated with human-induced and natural stress.

To test this concept, the U.S. Environmental Protection Agency (EPA), USDA Agricultural Research Service (ARS), Instituto del Medio Ambiente y el Desarrollo Sustentable del Estado de Sonora (IMADES), and the French Institute of Research and Development (IRD, formerly ORSTOM) have initiated a regional approach to assess ecological risk under the interagency Semi-Arid Land-Surface-Atmosphere (SALSA) Exit EPA Disclaimer program. SALSA is a collaborative research effort composed of international scientists committed to the study of land degradation processes in the Upper San Pedro Watershed (U.S./Mexico) using advanced space-based technologies. A key component of the SALSA framework enlists landscape ecology as a theoretical basis from which to assess cumulative exposure to stress at multiple spatial and temporal scales. This project has focused its research into developing a system of landscape composition and pattern indicators that can be used to estimate current status, trend, and changes in ecological and hydrological condition. Specifically, it is designed to determine ecosystem vulnerability relative to large-scale natural or human-induced disturbances using a system of landscape pattern metrics derived from remote sensing, spatial statistics, and geographic information systems technology. The project uses the database from the North American Landscape Characterization (NALC) project (EPA 1993) which incorporates Landsat Multi-Spectral Scanner (MSS) imagery from three dates across the U.S. and Mexico to establish environmental trends over nearly a 20-year period. In the case of the Upper San Pedro Watershed, MSS imagery from 5 June 1973, 10 June 1986, and 2 June 1992 has been remapped and projected to UTM coordinates at 60-m pixel resolution to generate land cover data. Additionally, an 8 June 1997 Landsat Thematic Mapper (TM) image has been added to the database and resampled and mapped at 60-meter resolution for comparison; all image dates have been classified to Formation (Biome) level in a 10-class land cover system, including urban (Table 1).

Databases of all relevant biophysical data, in addition to land cover, have been assembled to support the analysis, and these have been interfaced with ecological and hydrologic process models to assess the consequences of landscape change on sustainable resources.


The objectives of this project are to:

This research is focused on the development of an ability to measure and detect landscape change over a broad watershed area of concern. Subsequently, it provides a method to document changes and determine ecosystem vulnerabilities through the use of change detection and indicator development. The project provides the added benefits of developing methodologies to assess land cover accuracy for the NALC datasets and to develop multi-temporal and multi-scale process models which relate to important selected assessment endpoints such as erosion, flooding frequency, and stream water quality. This project will also provide a large primary spatial database which can be utilized for both future research and resource management, and thus serves as a prototype for other large-scale western landscape assessment projects. Collectively, the information will improve our ability to interpret landscape indicators as they relate to water resources and ecosystem resilience. The research will benefit a number of different organizations who are principally interested in evaluating present and past cumulative impacts to the watershed and are formulating alternative management strategies to sustain environmental health and economical viability into the future. Lastly, the research is anticipated to improve the ability and methodology of EPA to target and geographically prioritize locations for community-based environmental protection.


The current research is focused on the upper San Pedro River basin which originates in Sonora, Mexico and flows north into southeastern Arizona (Figure 1). The San Pedro River is an international basin with significantly different cross border legal and land use practices. The watershed embodies a variety of characteristics which make it an exceptional outdoor laboratory for addressing a large number of scientific questions in arid and semi-arid hydrology, ecology, meteorology, and the social and policy sciences. The Upper San Pedro Watershed represents a transition area between the Sonoran and Chihuahuan deserts and is internationally renown for its biodiversity. It supports the second highest land mammal diversity in the world and provides habitat for almost 400 bird species. The riparian zoSan Pedro Riparian National Conservation Areane in Arizona has been acquired by the U.S. Department of the Interior and has been assigned special land status as a National Conservation Area by Congress. Topography, climate, and vegetation vary across the watershed (Figure 2). Elevation ranges from 900 - 2,900 m and annual rainfall ranges from 300 to 750 mm. Biome types include desertscrub, grasslands, oak woodland-savannah, mesquite woodland, riparian forest, coniferous forest, and agriculture. The upper watershed encompasses an area of approximately 7,600 km2 (5,800 km2 in Arizona and 1,800 km2 in Sonora, Mexico).

All Landsat-MSS remote imagery has been acquired and pre-processed (i.e. triplicate scenes for three periods 1973, 1986, and 1992 have been coregistered and georeferenced to a 60 x 60 meter Universal Transverse Mercator ground coordinate grid). A derivative product (digital land cover map) has been developed for the image sets (including 1997 Landsat-TM) for the entire watershed. The landscape metric analyses have been reported in Kepner et al. 2000 and Kepner et al. 2002.

The land cover maps have been accuracy assessed by the University of Arizona (Arizona Remote Sensing Center). All spatial datasets (n>50) have been acquired and developed into a project database and the metadata have been compiled. Multi-scale hydrological process models which input landscape composition and pattern features have been tested using the Automated Geospatial Watershed Assessment (AGWA) tool jointly developed by EPA and ARS (EPA/600/R-02/046).

Figure 3 demonstrates four digital land cover maps derived from the Landsat satellite platforms (June 1973, 1986, 1992, and 1997) at 60m pixel resolution. Preliminary results indicate that extensive grassland and desertscrub areas with high connectivity are the most vulnerable biome type to encroachment of woody shrubs (mesquite). During the period of study, grasslands and desertscrub decreased by approximately 16% and 22%, respectively, as per cent total change. Xerophytic cover types, i.e. mesquite woodland, increased in overall extent by 388%. In 1997, grasslands remained the most extensive land cover type within the study area, however, human-dominated activity has lead to large increases in anthropogenic cover, e.g. urban, during the last 25 years.

Figure 4, Figure 5, Figure 6, and Figure 7 depict image differencing for grassland, desertscrub, mesquite woodland, and urban land cover types. The total extent by cover type is displayed for three periods (1973-1986, 1986-1992, 1992-1997) and their difference portrayed as simple gain/loss/no change maps.

Grasslands and desertscrub not only decreased in extent during the study period but also became more fragmented, i.e. the number of grassland and desertscrub patches increased by approximately 15% and 52%, respectively, and the average grassland and desertscrub patch sizes decreased approximately 27% and 49%, respectively (Table 2). These two dominant cover types not only dramatically changed in patch size metrics but they each became less connected over time. In stark contrast, mesquite woodland increased in connectivity by approximately 19% and the average patch size and number of patches increased 43% and 243%, respectively.

Urban cover has also increased during the study period. The majority of mesquite and urban gain duringUrbanization of San Pedro Valley (Arizona) the 25-year period was predominantly derived from grassland and desertscrub cover classes. Subsequently, grassland and desertscrub show a general trend in fragmentation and actual loss as they phase transform to mesquite woodland and newly urbanized areas (Figure 8).

The early focus of this exploratory study has been limited to the application of some simple landscape pattern metrics to demonstrate change detection methodology and landscape indicator development to assess watershed vulnerability in a semi-arid environment. Landscape analysis can be enhanced to accommodate more complex diagnostic statistics and can be applied to more detailed land cover maps with finer grain and greater classification detail (such as those derived from SPOT satellite imagery or fine-scale aerial photography). Application of landscape pattern analysis derived from classified remote imagery, e.g. Landsat-MSS and TM, provides an important new approach for environmental managers and policy-makers to measure and assess ecological condition and trend at many scales (e.g. site, watershed, region, nation). The process demonstrates a simple procedure to characterize ecological systems that are at potential risk from exposure to one or more stressors. Landscape analysis can evaluate environmental problems at multiple scales and can provide interpretive evaluations for various ecological, hydrological, or political assessment units. This approach integrates environmental information across all land cover types and provides an important new tool in which to evaluate land management and ecological change over time.


  Analytical Tools Interface for Landscape Assessments (ATtILA)
  Fact Sheets

Fact Sheet [PDF, 2 pp., 241 KB]

  Automated Geospatial Watershed Assessment (AGWA), EPA/600/R-02/046
  Fact Sheets
  Product Release [PDF, 1 p., 155 KB]
  Automated Geospatial Watershed Assessment: A GIS-Based Hydrologic Modeling Tool [PDF, 2 pp., 565KB]
  TMDL Fact Sheet [PDF, 2 pp., 693 KB]
  AGWA Brochure [PDF, 2 pp., 1.5 MB]

For more information on this subject:

See the Semi-Arid Land Surface-Atmosphere Program (SALSA) Exit EPA Disclaimer on the WWW.

Also see DOI U.S.-Mexico Field Coordinating Committee Exit EPA Disclaimer on the WWW.


ORD Home | NERL Home
ESD Home | LEB Home
Send questions or comments to ESD Info Desk. (Library-lv@epa.gov)

Local Navigation

Jump to main content.