The Use of MODIS-NDVI Data for Mapping Cropland across the Great Lakes Basin, USA

The goal of this work is to use the MODIS Normalized Difference Vegetation Index (NDVI) time-series data for cropland mapping of Great Lakes Basin (GLB). Three classification algorithms were examined to support cropland mapping, including a maximum likelihood, decision tree, and multi-layer perceptron neural network classifiers. The classification results were compared to the agricultural statistics from the NASS to assess relative performance. Results indicated that the neural network classifier produced best overall performance (SE = -2.46 and RMSE = 77.70), however, there were large variations in performances across the study region. A stratification of the GLB by ecoregion was used to improve the cropland classification performance.