Landsat–MODIS Assessment of Paddy Residue Burning in Northwestern India: District-Level Comparison of Burn-Affected Cropland During the 2021 Post-Monsoon Season
Keywords:
Paddy residue burning; Landsat; MODIS MCD64A1; dNBR; Burn-affected cropland; Remote sensing; Northwestern India.Abstract
Across northwestern India, a large part of the post-harvest agricultural fire activity is associated with paddy residue burning. The amount detected in regional satellite-based monitoring is not necessarily the same at different spatial resolutions, since smaller or fragmented burning areas may be represented differently. This study compares Landsat-derived probable burn-affected cropland with the MODIS MCD64A1 burnt-area product across seven validation districts in Punjab, Haryana and Rajasthan which focusing on the 2021 post-monsoon burning season within a comparison framework built from 2005 to 2025 data. Landsat Collection 2 surface reflectance imagery was processed using fixed pre- and post-burning seasonal composites to derive the Normalized Burn Ratio, Normalized Difference Vegetation Index, and their differenced forms, dNBR and dNDVI. A dNBR threshold of 0.65, established through long-term calibration against MODIS, was used to estimate probable burn-affected cropland for 2021. Landsat gave higher burn-area estimates than MODIS in all seven districts, despite this there is a difference between the two varied across the districts. Sangrur–Malerkotla showed the closest agreement, with Landsat and MODIS estimates of 2904.98 and 2795.59 km2 respectively, a ratio of 1.039, the largest difference was observed in Sonipat, where the Landsat estimate was more than six times higher than the MODIS value. The seven district-level observations produced a strong positive correlation between the two products (Pearson's r = 0.949, R² = 0.900, p = 0.0011), consistent with the pattern observed in the larger 2005 to 2025 clean-record dataset (n = 120, r = 0.850, R² = 0.722). The largest differences between the two products were seen in districts outside the main Punjab burning belt. The findings of the study suggest that MODIS-based estimates of burning in this region may be lower in areas where agricultural fields are smaller and more fragmented, and that Landsat's finer spatial resolution captures a greater extent of probable burning activity than the coarser MODIS product resolves.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 All articles published in this journal are lincensed under a

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.


