| In many cases, when classifying satellite imagery, training sites and sample data are not available on a yearly basis. Many locations might only have complete, quality ground data for a single year over a period of a decade or more. Therefore, it would be beneficial if accurate training data from a single year could be applied to other years.;The objectives of this research were to: (1) utilize time-series MODIS 250m NDVI data to identify and map unique crop types for the state of Kansas and the surrounding Kansas River watershed and (2) test the level of accuracy when conducting cross-year classifications by applying a library of NDVI time-series curves to imagery from other years.;MODIS 250m NDVI data were used to classify seven unique crop cover types for 2005, including winter wheat, corn, soybeans, sorghum, alfalfa, fallow, and double crop. The classified maps' patterns were consistent with the cropping practices of the study area and an overall accuracy of 82% was achieved.;MODIS 250m time-series NDVI data along with Common Land Unit (CLU) and Farm Service Agency (FSA) training and validation data from 2001 and 2005 were used to conduct the cross-year classifications. Overall accuracies were found to be between 68% (2001) and 74% (2005). The general patterns of the classified maps were consistent with the state's cropping practices. The relatively low accuracy levels are likely due to variations in climatic events and farming practices between the two years. |
