For the purposes of this report, the detailed categories for land cover as defined in either the GIRAS or APES data sets were consolidated into seven general classifications--residential, urban, agriculture, forests, lakes/reservoirs, forested wetlands, and transitional areas. For representation purposes, these seven categories were further grouped into three main land covers--developed (residential and urban), agricultural, and forested/other (forests, lakes/reservoirs, forested wetlands, and transitional areas).
Due to the differences in photo interpretation and land-use classification schemes, discrepancies exist in the categorization of land cover between the two data bases. For example, in the GIRAS data base, land-cover features are delineated by polygons with a minimum 4-hectare size (Mitchell and others, 1977). Thus, a small watershed with fragmented numerous land uses may be misrepresented in classification due to the limited extent of land cover expressed in the imagery. The urban classification category, as defined in the APES data base, may be the least accurate in representation of actual land cover (Dodd and others, 1992). Additionally, areas unable to be classified have been integrated into a "transitional area" category which may increase the disparity between the two land-use coverages. To minimize errors in the urban and residential land-cover categories, 1975 and 1988 data bases were overlain with population coverages from the 1980 and 1990 censuses, respectively. Land-cover polygons with census-derived populations exceeding 1,000/mi2 were reclassified as urban regardless of the original land-cover classification. Polygons with populations exceeding 500/mi2 but less than 1,000/mi2 were classified as residential. The unlikely small decrease in developed area from 1975 to 1988 shown for the Ellerbe Creek watershed (fig. 2B) is probably a result of this type of error.
Return to Water-Quality Trends for Streams and Reservoirs in the Research Triangle Area of North Carolina, 1983-95