A clear call for research on land use and land cover change resulted from the National Research Council (NRC) response to a National Science Foundation (NSF) request to identify the "Grand Challenges in Environmental Sciences" (NRC, 2001). An interdisciplinary committee was asked to determine the most important research challenges over the next 20 to 30 years within the context of environmental problems. One of eight grand challenges was Land Use Dynamics, which calls for the development of a comprehensive understanding of changes in land use and land cover that are critical to biogeochemical cycling, ecosystem functioning and services, and human welfare. The report concluded that “…improved information on and understanding of land use and land cover dynamics are essential for society to respond effectively to environmental changes and to manage human impacts on environmental systems" (NRC, 2001).
Two additional NRC reports emphasize the importance of land use and land cover change research. A 1999 report on Measures of Environmental Performance and Ecosystem Condition called for investigations of the complex relationships between humans and the environment and emphasized data collection and monitoring of both ecosystem processes and land use and land cover change (NRC, 1999). Another NRC report titled Ecological Indicators for the Nation declared that the largest ecological changes caused by humans result from land use (NRC, 2000). Because these changes affect the ability of ecosystems to provide the goods and services that society depends on, an assessment of land cover change is needed to understand the status of the Nation's biological resources.
While a great deal has been written regarding change-detection techniques using remotely sensed data, very little guidance exists for addressing large-area change detection (Dobson and Bright, 1994).
Large-area change detection has generally relied on low-resolution sensors, such as the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR), to provide information on general changes in vegetation indices or similar measures (Tucker et a1., 1986; Helldenand Eklundh, 1988; Lambin and Strahler, 1994). The spatial resolution of such sensors, however, makes it difficult to identify and quantify the types of fine-scale land cover changes that are often associated with anthropogenic change. The use of moderate-resolution imagery (i.e., 30-meters), such as Landsat TM data, makes this task much more feasible. However, wall-to-wall change detection using moderate- to high-resolution imagery for large areas presents stiff challenges with respect to accuracy, time, processing loads, and budgets.
Spectral data recorded by remote sensing instruments can provide information on land cover conversions and on changes in condition, but it is generally not a consistent indicator of land cover change. Land cover transitions often have very small changes in spectral response and may not be readily identifiable. Interpretation of tone, texture, shape, size, and pattern can all help to identify land cover change, but these elements are disregarded in many automated change analysis studies.
The most straightforward technique for detecting change is the comparison of land cover classifications from two dates. The use of independently produced classifications has the advantage of compensating for varied atmospheric and phenological conditions between dates, or even the use of different sensors between dates, because each classification is independently produced and mapped to a common thematic reference. The method has been criticized however, because it tends to compound any errors that may have occurred in the two initial classifications (Gordon, 1980; Stow et al., 1980; Singh, 1989). The procedure has been successfully used in various land cover change investigations, including assessing deforestation (Massart et al., 1995), urbanization (Dimyati et al., 1996), sand dune changes (Kumar et al., 1993), and the conversion of semi natural vegetation to agricultural grassland (Wilcock and Cooper, 1993).
Simultaneous analysis techniques, including image differencing, ratioing, principal components analysis (PCA), and change vector analysis, are common change analysis approaches. Image differencing (i.e., subtraction between georegistered images (raw or transformed) from two dates) is probably the most widely used approach (Weismiller et al., 1977; Vogelmann, 1988). Image ratioing (Howarth and Wickware, 1981) and PCA (Bryne et al., 1980; Ribed and Lopez, 1995) have also been widely used. Sohl (1999) successfully used change vector analysis to document land cover change in the United Arab Emirates. Although often effective at identifying areas of spectral change, these techniques typically result in the creation of a simple, binary change masks.