Development of land-use map for Salt River basin using satellite imagery

Abstract

The Salt River Basin in northeastern Missouri is the major water supply source for people in the region. There is a need to build a long term simulation and to validate the long term trend of pollutant concentrations in the basin. Remote sensing is a potential resource to support land-use evaluation. It provides the capability to evaluate dynamic landscape pattern and observe changes and trends across a large scale pattern through time. This research applied Landsat 5 Thematic Mapper (TM) data to build crop type maps for the Salt River Basin for a two-year period (2005 and 2006) useful in evaluating land-use management with crop phenological profiles analyzed with MODIS NDVI. The results showed that crop type classification in 2005 had a greater accuracy than that of 2006. The primary reason was that there were more in-season Landsat data available in 2005 than in 2006. This method was efficient in classifying crops when there was lack of cloud-free images during the growing season as indicated by an overall accuracy of 90% in 2006. Combining this method with statistical analysis using additional ground-truth samples for building the NDVI profile has potential as a robust technique for crop classification in cloud-free-limited Landsat data application useful in other studies.