Linear mixture modelling applied to AVHRR data for crop area estimation

Abstract A technique for estimating crop coverage using linear mixture modelling of multi-temporal Advanced Very High Resolution Radiometer (AVHRR) data is presented for a study area in northern Greece. This paper identifies some of the problems associated with using satellite sensor data with coarse spatial resolution for crop area estimation. Using satellite sensor imagery with a high spatial resolution to extrapolate ground measurements to AVHRR scales, the paper shows how the mixture model can be applied to AVHRR data in a mixed agricultural system. Crop areas are estimated to an average accuracy of 89 percent on regional scale using this technique. The results show that this linear mixture modelling has potential for operational crop area monitoring on a regional basis.