The creation of a land use map of the Portland, Oregon region from LANDSAT has proven to be a useful demonstration that helped substantiate the utility of satellite thematic mapping for regional planning purposes. The tabulation of land use acreage by traffic zones and census tracts, with the application of Image Based Information System (IBIS) software, has permitted a more systematic verification of the LANDSAT classification. Subsequent tabular data base constructions merged socioeconomic and environmental data with the LANDSAT-derived land use statistics. The application of IBIS software to the data bases permitted the cross-tabulation of population statistics and modeling and portrayal of per capita air pollution impact.

The development and system characteristics of an image based information system (IBIS) that makes use of digital image processing techniques has been described by Bryant and Zobrist in June 1976. A brief summary of the approach is described using Portland case as an example.

The generation of land use statistics by census tracts and traffic zones permitted the direct comparisons of acreage estimates derived from LANDSAT with those obtained from conventional air photo interpretation and field surveys. The systematic errors associated with the Bayesian classifier as it was applied to different LANDSAT frame dates could be noted for different portions of the scene. Thus, it was possible to note the degree of misclassification in the various suburban, industrialized, and central business district areas of the region.

The merging of the LANDSAT-derived tabular data sets with socioeconomic data was accomplished using conventional sort-merge routines. IBIS-based applications included several spatially-dependent manipulations of both kinds of data. In the first instance population statistics by census tract were allocated to traffic zones in accordance with the LANDSAT- derived distribution of residential land use. Similarly, it was possible to use residential and commercial land use acreage by traffic zone as coefficients of trip generation to map potential automobile trip generation. Finally, it \&as possible to model air pollution impacts through the pixel-for-pixel cross correlation of land use and socioeconomic data collected by census tract and environmental data collected by two kilometer grid cells and map the results as a contoured surface.

Date of this Version