An Integrated Remote Sensing Approach For Regional Agro-statistics And Land Monitoring

A set of nine satellite images of six different recording dates were merged together after radiometric recalibration and geometrical correction to the Belgian Lambert Cartographic ProJection. This multitemporal dataset was classified with a decision tree incorporating single channel level-slicing, diFferent vegetation-indices and a minimum distance classiFication algorithm, all based on the recalibrated digital values. Fourteen terrain-classes could be distinguished including ten crops, deciduousand coniferous forest, waterbodies and urbanisation. The classification proved to be in accordance with the data gathered by the National Institute of Statistics.