2017 Volume 33 Issue 2 Pages 27-36
In Indonesia, agriculture is one of the main economic sectors and source of national income for the country. Therefore, precise agricultural mapping and information are very important for both the national and regional administration. Satellite remote sensing is viewed as the most effective tool for identifying the agriculture croplands for wide area. However, cloud coverage in tropical region limits the use of conventional optical remote sensing. For this reason, the Synthetic Aperture Radar (SAR) application has advantages for monitoring agriculture in tropical region. This study applied ALOS PALSAR imagery with full polarization mode for classifying the agricultural croplands. A study area of about 50km2 located in Central Java was chosen as it consists of the mixed and complex agriculture croplands common to Java Island. The backscatter intensity of the full polarization image was processed and combined to extract the most suitable parameter. Two kinds of polarimetric decomposition were also applied and analyzed for a better understanding of the polarimetric scattering mechanisms of agricultural croplands. Overall, the integration of backscatter intensities (HH,HV,VH,VV, and HH+HV) and Freeman and Durden polarimetric decomposition showed the best accuracy for classifying the agriculture croplands with an overall accuracy 74.11% and a kappa coefficient of 0.62. This fact implies that the integration of both backscatter intensity and polarimetric decomposition methods was able to compensate for the weakness of each component in discriminating the complex agricultural croplands.
