Appendix I - Land Use and Land Cover Assessment

These maps provided an overview of land use within the watershed. Subbasins were delineated on 7.5 minute quadrangle topographic maps by the U.S. Geological Survey staff and digitized using ARC/INFO GIS software. Methods follow those described by Rains and Latham (1993). Satellite imagery was selected as the source for determining existing land use and land cover data. Existing land use and land cover were mapped using 1993 LANDSAT Thematic Mapper 30 meter resolution multispectral satellite imagery. Satellite image analysis was further refined using secondary data, including low altitude air photos and ground truthing. The classified satellite image raster files were converted to ARC/INFO vector files. Existing land use and land cover maps were geo-referenced to 7.5 minute quadrangles and merged to produce seamless basin-wide coverage.

Existing land use and land cover were classified using the Florida Land Use, Cover, and Forms Classification System (FLUCCS) developed by the Florida Department of Transportation (1985). Land use and vegetation cover were classified hierarchically, with each successive level including land information of increasing specificity. Level II (and level III where possible) of the FLUCCS hierarchy was used. Although FLUCCS categorizes land cover as either uplands or wetlands, many of the "upland" forests classes were found in low, wet areas and contained approximately equal portions of upland and wetland vegetation. For the purposes of this report the land use and land cover data has been aggregated. More detailed information is available on request.

Sites were selected for class verification base on accessibility. Much of the land within the basin either private property or not accessible by road. Accuracy of the existing land use and land cover classification was assessed by comparing predicted land use classes to observed classes, determined by air photo interpretation and/or field verification.

A minimum map accuracy of 85 percent, with a 90 percent to 95 percent confidence interval, is standard for most natural resource land use studies. It was beyond of the scope of the project to perform an accuracy assessment which could verify this standard. However, based on results the observed during the field verification performed and several similar studies completed by the District, it is anticipated these standards would be verified by further assessment.

Use of satellite imagery or aerial photography and maps for planning or regulatory purposes varies for regional, local, and site-specific endeavors. In this study, land use and land cover mapping was accomplished using 30 square meter satellite imagery with plus or minus one pixel resolution (30 square meters). While satellite imagery is an exceptionally good source of regional land cover and land use data, it may be less satisfactory for urban areas and linear features due to resolution limitations. The resolution limits the ability to differentiate among some urban features using the imagery.

Existing land use maps identified many but not all streams in the St. Marks and Wakulla Rivers watershed. Linear features such as roads, streams, and stream-associated wetlands are particularly difficult to distinguish using satellite imagery classification. To register in satellite imagery, a linear feature must dominate the 30 square meter pixel. If a 15 meter wide stream and associated wetland vegetation bounded by 20 square meters of forest occurs at the pixel edge, only the forest may occur in the imagery due to its areal extent. Satellite imagery detects linear features in areas with high spectral contrast (such as a dirt road bounded by forest or bridge bounded by water), but tends to miss linear features with low spectral contrast (such as a stream bounded by a marsh). As a result, linear features often appear discontinuous when mapped using satellite imagery.

Recently, the water management districts (WMDs), FDEP and other regional and state agencies together purchased statewide Landsat satellite imagery. St. Johns, South Florida, Southwest Florida and Suwannee WMDs and local governments have also shared the cost of Photogrammetric mapping. The resulting purchases were made at a lower cost than previously budgeted by these agencies for only a portion of the state. Cost-sharing among state, regional and local agencies for land use and land cover mapping could result in better data and maps at a lower cost.

As part of the ongoing District Water Management Plan efforts, WMDs and FDEP agreed to use the 1985 FLUCCS as the primary classification system in order to ensure compatible land use and land cover data on a statewide basis. The districts and FDEP also agreed to recommend that DCA develop standards for the comprehensive plan future land use map to ensure greater consistency of base maps, scales, and cartographic standards. As land use and land cover data via satellite imagery become increasingly available, cost-sharing by state, regional and local agencies of land use change detection mapping and analysis would provide a good basis for monitoring comprehensive plan future land use implementation and estimating NPS loadings.