Mapping of Agricultural Activities Using Multi-temporal Asar Envisat Data

Space borne radar data facilitates continuous monitoring of almost any location on the earth (limitations in polar zones) at quite low costs. Almost weather-independent operation of radar systems enables a reliable and continuous record of data from earth’s surface. In the framework of an ESA pilot project (AO335), ENVISAT polarimetric SAR data of year 2004 are examined for their usefulness in environmental monitoring within a drinking water protection area, north east of the city Hanover in Germany. This is done by using ENVISAT ASAR images together with GIS information like topographic maps, orthophotos and also ground surveys. Because of only 2 polarisations of ASAR, with a coherent response of different vegetation types and the high variance of pixel values, the results from classification approaches using monotemporal images are unsatisfactory. Our experiments and the experience of other authors as well as the knowledge about crop phenology led to a multi-temporal classification approach improving the classical methods. In multi-temporal classification, images from different dates, which cover the phenological period of desired crops, are treated as bands of a multi-temporal image. The feasibility and accuracy of this multitemporal approach is evaluated using a pixel based approach. The benefit of some pixel based classification rules, the influence of some speckle filters on overall accuracy and the importance of adaptation to phenological period of crops are tested for this approach.