Classification of multipolarisation radar images in agricultural areas

The objective of this research is to study the contribution of multipolarisation airborne radar data to crop discrimination. An unsupervised classification algorithm and a supervised method based on maximum likelihood were used and compared for agricultural applications. The experimental area, in Southern Ontario, Canada, was chosen because it has been the site of extensive inventory in relation to the application of VIR and radar data for agricultural uses. C-HH, C-VV and C-HV data for 10 July 1990 were used for the study. At this time, crop development allows optimal separability between crops. Results show that multipolarized radar data offer an adequate tool for crop identification. Correct classification rates of 83% and 79% were obtained for supervised and unsupervised methods respectively. Comparison of the two methods reveals that the performance of the unsupervised classification is similar to that of the supervised classification. This is a promising result if the authors take into consideration the fact that the unsupervised classification eliminates tedious effort for data collection and that it makes mole efficient use of computer time in the training stage