2005 Salt Lake City Annual Meeting (October 16–19, 2005)

Paper No. 29
Presentation Time: 1:30 PM-5:30 PM

ANALYSIS OF ECOSISLA, ECOSYSTEM, AND COUNTRY LAND COVER CLASSIFICATION IN CENTRAL AMERICA


DANZ, Julia M., Geology, Beloit College, 700 College St, Beloit, WI 53511, danzj@stu.beloit.edu

Satellite images have provided an incredibly efficient way of analyzing land cover over large areas, but these are often not 100 percent accurate for several reasons. Comparing different land cover classifications helps in recognizing these factors by giving a better understanding of the accuracy in these classifications and identifying areas for improvement. There were three land cover classifications for Central America that were compared in this study. These were Ecosisla, Ecosystem, and specific country land cover classifications, all of which have a different number of polygons in the same area, creating different levels of detail. The classifications for each Central American Country were the most detailed, followed by Ecosystem, and then Ecosisla. The country classifications were created by each individual country that had their own systems for categorizing land cover types, and these were narrowed down into the seven categories found in Ecosisla and Ecosystem – forestland, cropland, grassland, wetlands, settlements, other land, and no data. To determine how accurate these land cover classifications are, they were intersected together for comparison. The country classifications were intersected with Ecosisla, and with Ecosystem, and Ecosisla was intersected with Ecosystem. The total areas of the different land cover types where put into a matrix to find which categories agreed and the percentage of agreement for the category of each map. By laying out the categories, the reasons low accuracy was identified. Some of the major causes of disagreement are country land cover classifications have more polygons than Ecosystem or Ecosisla, land cover categories that have a small total area are more likely to have a lower accuracy since the smaller amount of data makes them more susceptible to errors, and Land cover classifications that consist of many small polygons, such as the country maps, are far more susceptible to disagreement if the maps don't align. These can be used as signs for potential disagreement, as well as factors to consider when working with land cover classifications.