| Abstrakt: |
Agriculture plays crucial role for developing the economic status of any country. As the population is increasing day by day, the demand for food materials is also increasing. For the storage, import, export, pricing, etc., of food materials, there is need of timely production forecast of the crops. To get timely and accurate production forecast, there is requirement of continuous monitoring of the agricultural crops. The demand of early forecast creates the opportunity of using remote sensing to provide timely and accurate crop forecast using satellite-based technology. Mustard is an important oilseed and cash crop grown during the rabi season under irrigated or assured water conditions. The objective of this study is to discriminate the mustard crop from other rabi season crops by using the temporal MSI and SAR data of Haryana state of India. Based on crop spectral profile of MSI data and temporal backscatter profile of SAR data, classification has been done by random forest classification technique and the LULC was segregated into the mustard and other classes to generate the mustard crop classified mask. In the study, NDVI values of the crop derived from multidate MSI data are compared with backscattering values obtained from multi-temporal SAR data. The R square between NDVI and SAR backscatter is 0.907 which is showing positive correlation between both. The classification accuracy for the mustard crop was found to be 95% and 91.66% using SAR and MSI data, respectively. The present study suggests the potential of multi-date Sentinel-1A VH polarized SAR data for differentiating the mustard crop from the other associated rabi season crops using machine learning approach. |
Full text from SpringerLink