CIESIN Thematic Guides

Image Interpretation

Lillesand and Kiefer (1987) discuss the considerable body of knowledge associated with airphoto interpretation that has developed over the past six decades with the evolution of aerial photography throughout the book Remote Sensing and Image Interpretation. Some of the techniques are applicable to satellite images in varying degrees, depending on the level of spectral information and spatial resolution they provide.

Image stratification represents a fundamental image-interpretation technique that enables the use of satellite images as a spatial frame of reference. For human dimensions studies, the objectives of stratification are to delineate relatively homogeneous areas of landscape to 1) regionalize the patterns of occurrence associated with selected human activities; 2) direct subsequent efforts of qualitative evaluation and/or airborne or field data acquisition to appropriate regions; and 3) reduce the variance associated with quantitative estimates of selected landscape variables associated with human activities.Figure 1 illustrates stratification of a satellite image for regionalizing different densities of urban population.

Frequently, the joint use of two or more levels of satellite image data will improve the information interpreted from images. For example, stratification may be implemented in a nested fashion on progressively higher resolution satellite images. One might use image data of coarse spatial resolution (such as 1-km Advanced Very High Resolution Radiometer data) to stratify a large region by delineating major natural land cover formations, areas of anthropogenic land use (such as urban and various agricultural regions), and any evidence of land-cover change. At subregional and local scales, higher resolution satellite image data or airphotos will enable subsequent stratification of homogeneous areas according to characteristics such as vegetation type and density, field size, cropping patterns, or types and densities of disturbances (such as cutting, burning, and harvesting in forested regions). Roller and Colwell (1986) discuss several examples of the use of coarse-resolution image stratification for improving the inventory of environmental resources in "Coarse-resolution Satellite Data for Ecological Surveys."

Interpretation techniques also include the detection, identification, and measurement of specific features. Examples include 1) location and delineation of selected cultural features and evidence of human activity, such as hamlets, agricultural fields, or forest cutting; 2) enumeration and rank ordering of the significance of such features according to characteristics such as relative size or proximity to each other; and 3) area determination through direct image measurement or sampling and estimation techniques. Figure 2 illustrates a satellite image of the Western Sudan that allows extensive interpretation of human activities.

The joint interpretation of satellite images and relevant ancillary data can greatly aid the extraction and evaluation of information. Images can be overlain with social structure strata, such as political or census boundaries or cultural information, to provide meaningful context for interpretation and analysis of human-dimensions issues. As an example, the merging of census boundaries and population attributes with Landsat images will enable population data to be evaluated relative to observable human disturbances or other elements of the physical environment, a type of analysis not generally feasible with just population data.

Ancillary data can also reduce the physical workload associated with satellite image interpretation. The task of interpreting current land use can be facilitated by using previously compiled map or airphoto information. Such information may have been produced in developing regions of the world as part of regional and national programs addressing agricultural or commercial development, population resettlement, or crisis management. If previous maps or airphoto interpretations of land cover and/or land use exist, merging these with current images can provide a basis for interpreting land use through comparison. In this way, current land use is interpreted through an update process rather than an independent interpretation process. Figure 3 illustrates an example of such an update process.


Figures 1 - 3


A final step in the image interpretation process may involve digitizing the results. Digitizing enables capturing the information in computerized spatial database format for subsequent merging and joint analysis with other spatially formatted data in a geographic information system (GIS). For example, Skole and Tucker (1993) report results of interpreting deforestation throughout the Brazilian Amazon Basin on Landsat satellite imagery and digitizing the results for spatial analysis of tropical deforestation and habitat fragmentation in a GIS in "Tropical Deforestation and Habitat Fragmentation in the Amazon."