South America Land Use and Land Cover Assessment and Preliminary Analysis of Their Impacts on Regional Atmospheric Modeling Studies

Data provided by two important sources of information on land use and land cover (LULC), MODIS-2009 and GLOBCOVER-2009, were analyzed for South America in order to assess differences related to their application in numerical modeling studies. Even though on a South American basis, the two databases showed a Pearson correlation coefficient above 85%, on a regional analysis, the correlation stayed within the range of 0%-100%, depending on the territorial unit analyzed. Significant differences were observed in most of the land cover classes, with only forested areas presenting a good level of agreement. In terms of territorial units, only areas in the Amazon region, where forest cover is predominant, showed significant correlation levels. Crops and urban classes presented the greatest differences between the two analyzed files. Results of meteorological simulations indicated that such observed discrepancies are able to cause strong impacts on modeling scenarios and important bias on simulated variables, being a crucial feature for weather and climate forecast and diagnostic.