Classification of Northern Thai Rice Varieties Using Random Forest (RF) and Support Vector Machine (SVM) on Google Earth Engine with Sentinel Imagery: A Case Study in Buak Khang Subdistrict, San Kamphaeng District, Chiang Mai Province

Abstract

Rice is a crucial agricultural product for Thailand's economy, as the majority of the country's agricultural sector primarily cultivates rice for both domestic consumption and international demand. This research focuses on land use analysis, specifically on rice cultivation areas, and the classification of different rice varieties. The study significantly contributes to the achievement of the Sustainable Development Goals (SDGs) concerning food security. This research utilizes Sentinel-2 Spectral Instrument, Level 2A, and Sentinel-1 polarization VV and VH satellite images from 2023, covering the planting season (June to November). The data processing and analysis are conducted on the cloud platform Google Earth Engine (GEE), employing Support Vector Machine (SVM) and Random Forest (RF) classification methods for both land use analysis and rice variety classification. The analysis results indicate that the RF classification method has higher accuracy than the SVM method. Specifically, for land use analysis, the RF and SVM classification methods achieved accuracy values of 0.91 and 0.87, and kappa values of 0.89 and 0.85, respectively. For rice variety classification, the RF and SVM methods achieved accuracy values of 0.88 and 0.83, and kappa values of 0.73 and 0.61, respectively.