Geographia Technica, Vol 19, Issue 2, 2024, pp. 46-56
ASSESSING AGRICULTURAL BURNED AREAS USING dNBR INDEX FROM SENTINEL-2 SELTLLITE DATA IN CHIANG MAI, THAILAND
Ratchaphon SAMPHUTTHANONT 
ABSTRACT: This study conducted an assessment of agricultural burned areas using the dNBR index from Sentinel-2 satellite data, HARMONIZED collection, in Mae Rim District, Chiang Mai Province, over 5 years from 2019 to 2023. A total of 118 satellite datasets, before and after burning, were used to analyze the severity levels. It is considered severe once the Moderate Low Severity level is above 0.27. Error correction employed scene classification data, derived from the European Space Agency's area classification algorithms, and the NDWI index was used to exclude water-covered areas. Accuracy verification through an Error Matrix was conducted at 73 survey points with a 95% confidence level, adhering to the principles of statistical probability. The overall accuracy of the burned area classification was 82.19%. Agricultural burned areas in the study area were predominantly found mostly on flat terrain. in the eastern direction. In general, there was a significant increasing trend in burning, especially in the latest year 2023. Changes in the distribution burning month distribution were observed; in 2019-2022 burning was more prevalent in May, while in 2023, it shifted to April. This study successfully detected rice field burning; a small-scale, low fuel load, and low temperature burning, which the satellite hotspot data could not detect such burning. The results provide valuable information to promote the creative reduction of burning in communities by utilizing post-harvest agricultural residue, demonstrating the timely and appropriate application of tools and data for societal benefits.
Keywords: Agricultural Burned Area, Remote Sensing, Difference Normalized Burn Ratio (dNBR), Sentinel-2
Full article here