Mapping Rice in Viet Nam: Estimating rice area using remote sensing and area frame survey methods in An Giang, Viet Nam
Conference
10th International Conference on Agricultural Statistics
Format: CPS Paper - ICAS 2026
Keywords: "model-assisted, machine learning, remote sensing, rice area mapping, sampling frame, vietnam
Abstract
Rice cultivation plays a key role in Viet Nam’s agricultural sector, contributing significantly to the country’s food security and economic livelihoods. In An Giang province, one of the country’s main rice-producing regions, accurate and timely mapping of rice cultivation areas is essential for economic planning and disaster monitoring. This study aimed to map the rice cultivation area in An Giang province during the summer-autumn season of 2024, using remote sensing and machine learning methods, the Japan Aerospace Exploration Agency’s International Asian Harvest Monitoring System for Rice (INAHOR), and grid-based area sampling techniques to design an effective ground truth sample and compare rice area estimates.
Our remote sensing-based jackknife estimate of the rice cultivation area in An Giang province for the summer-autumn season is 2,500.10 square km2 (coefficient of variation [CV] 0.19%), with an overall accuracy of 95.3%. This is consistent with area estimates derived from the area sampling approach using the expansion estimator (2,517.63 sq km2, CV 8.04%), the ratio estimator (2,444.51 sq km2, CV 5.02%) and the adjusted ratio estimator (2,501.06 sq km2, CV 0.3%) for comparison. Our estimates further align with the administrative report of the National Statistics Office of Viet Nam, which indicated an area of 2,284.77 km2, approximately 9% lower than our remote sensing-based estimate. The study outcomes demonstrate the effectiveness of a more objective, design-based approach to remote sensing for rice area estimation.