158 
remainder of the maps for the State. The task of 
digitizing and editing these extensive polygon files was 
completed by the Graphics Division in the Spring of 1982. 
Subsequently, the GIRAS polygon files have been processed 
into 63.5 meter grid cells that correspond geographically 
to the Landsat derived data base. 
COMPARISON 
The completion of the two independently derived land cover 
classifications provides an excellent base for comparative 
analysis of the different remote sensing methodologies. 
Although there is some difference in the dates on which 
the various imagery were processed, the data were all 
collected within a seven year period between 1972 and 1979, 
with 1977 being a median date for both. Similarly, 
although one procedure started with rasterized data and the 
other with polygons, great care has been taken to insure 
that the relevant data cells are both the same size and 
geographically referenced to the same grid. 
A rigorous analytical comparison of the two data sets has 
been delayed until two major projects are completed. Thus 
far, preliminary visual analysis of various output products 
indicate that some radical differences exist between the 
two files. Both of the data sets have been divided into 
various subsets and samples in order to facilitate their 
display on the image processing system. For example, the 
entire rectangular file of one acre cells contains 5722 
rows and 7255 columns. Less than one per cent of these 
more than 41 million cells can be displayed at one time on 
the monitor with a resolution of 512 by 512 pixels. There 
fore most of the state-wide comparisons have been made by 
sampling every fourteenth row and column. With the image 
processing system it is possible to easily change colors, 
pan, zoom and annotate an image. Excellent color slides 
and 8" by 10" Polariod prints can be made with a Matrix 
color camera system. Grey scale hardcopy is made on a 
22 inch electrostatic plotter. 
Based on the visual inspection, it appears as if there is 
fairly good agreement between the two classifications for 
all categories except forests and wetlands. The urban 
areas were deliberately omitted from the Landsat classifi 
cations. The USGS classification of urban land consists of 
seven level II categories of land use, including transpor 
tation which delineates, among other features, the Inter 
state Highway System (Fig. 3). While there is general 
agreement about the location of urban land areas the USGS 
file is more extensive and detailed. 
Both water bodies and agricultural land use display a 
considerable amount of agreement in the two data sets. Due 
to the fact that the two remote sensing procedures utilized 
can easily differentiate water bodies, the only differences 
probably correspond to small ponds that could be detected 
from the Landsat data but were omitted from the USGS 
classification, as they did not meet the minimum mapping 
size requirement of ten acres. Since the USGS