Estimation of Boro rice area in Bangladesh using Sentinel-2 imagery and machine learning algorithms

Climate Change and Agricultural Adaptation in Bangladesh

Abstract

Rice is the primary food crop in Bangladesh, and a significant portion of agricultural land is dedicated to its cultivation. Annually, approximately 36.87 million tons of rice are produced from 11.54 million hectares of land (BBS, 2022). Among the three rice seasons in Bangladesh, namely aus, aman, and boro, the winter rice boro production holds the highest percentage. Accurate delineation of boro rice-growing area is crucial for estimation of total production, which plays a vital role in policy planning and decision-making. Rice fields in Bangladesh are fragmented into smaller plots, emphasizing the importance of high-resolution, cloud-free satellite images for precise delineation of rice areas. In this context, a comparative study was conducted to estimate boro rice area in Bangladesh using high-resolution (10-meter) Sentinel-2 data, aiming to overcome the challenges posed by fragmented land. Machine Learning (ML) based supervised classification algorithms namely Decision Tree, k-Nearest Neighbor (k-NN), Random Forest (RF) and Support Vector Machine (SVM) were employed on Sentinel-2 images of the study area to identify rice fields. The study's findings are expected to contribute to the development of boro rice area estimation, predict yield and productivity system in Bangladesh, ultimately enhancing food security and the livelihood of farmers.  

Bangladesh J. Agri. 2025, 50(1): 72-82