APPLICATION OF FINE QUAD-POLARIZATION RADARSAT-2 DATA AND SUPPORT VECTOR MACHINES TO CLASSIFY OFFSEASON RICE PADDIES

Abstract

In monsoon areas in Asia, most seasonal rice paddies are destroyed annually by flooding.
To secure rice production, farmers in these areas grow off-season rice crops, which can be produced only
with irrigated water. However, due to prevailing drought conditions, the cultivation of off-season rice crops
cannot commence simultaneously. The availability of irrigated water is the critical factor for farmers in
deciding when to begin off-season cultivation. To assess off-season rice production, a map of the cultivated
area is required. Using multi-polarization/multi-temporal RADARSAT-2 data, this study aims to estimate
and map the off-season rice paddies. The potential of using support vector machine classification to map
off-season rice paddies with C-band SAR data was also established. Multi-temporal fine quad-polarization
RADARSAT-2 data can be effectively used to estimate off-season rice cultivated areas.