Mapping Paddy Rice Fields Using Landsat and Sentinel Radar Images in Urban Areas for Agriculture Planning

Abstract

     This research develops a new method based on spectral indices and random forest classifier to detect paddy rice areas and then assess their distributions regarding to urban areas. The classification will be conducted on Landsat OLI images and Landsat OLI/Sentinel 1 SAR data. Consequently, developing a new spectral index by analyzing the relative importance of Landsat bands will be calculated by the random forest. The new spectral index has improved depending on the most three important bands, then two additional indices including the normalized difference vegetation index (NDVI), and standardized difference built-up index (NDBI) have been used to extract paddy rice fields from the data. Several experiments being conducted to analyze and understand the strengths and weakness of the proposed new method. This research shows that spectral indices are easy and accurate tool for rapid mapping of paddy rice fields in complicated environment where urban features are dominated. The outcomes of this research could help mapping and decision makers to progress their productivity and strategic plans for better management of rice fields.