Utilization of Sentinel-1 and Sentinel-2 Time-Series Data for Mapping Paddy Fields Changes in Klaten Regency, Indonesia

Abstract

Paddy fields area has a strategic function as the main food provider for the majority of Indonesia's population. Based on the fixed area of paddy fields from 2013 to 2019, there have been three significant changes in the national paddy fields area. Changes in paddy fields into other land use occurred massively, especially in urban and semi-urban areas. The utilization of remote sensing can be one of the methods in detecting changes in paddy fields area temporally. The purpose of this study was to calculate changes in the area and its distribution of paddy fields in Klaten Regency from 2016 to 2020 using multi-sensor remote sensing data. The method used in this study was detecting paddy fields area temporally in 2016, 2018, and 2020 adopting machine learning approach Random Forest classification using Sentinel-1 and Sentine1-2 imagery. The results of the analysis of paddy fields area with Random Forest method using Sentine1-2 indicated an average test accuracy value of 95% compared to Sentinel-1 analysis with an average test accuracy value of 92%. The results of the analysis of the changes in paddy fields area in Klaten Regency using Sentinel-1 showed that in 2016 there were 30,480 hectares of paddy fields area and in 2020 it decreased to 29,588 hectares. Furthermore, the results of the analysis paddy fields area using Sentine1-2 in 2016 was 31,683 hectares and decreased to 31,196 hectares in 2020. These results demonstrated an under-estimate in the calculation of the results of paddy fields area analyzed by Sentinel-1 and Sentine1-2 compared to the officially fixed area of paddy fields released by ATR/BPN.