Optimising crop-residue burning PM2.5 emissions over the Indo-Gangetic Plain using inverse modelling

The Indo-Gangetic Plain (IGP)—one of the most polluted regions globally—experiences intense seasonal crop-residue burning, yet its contribution to ambient PM₂.₅ remains poorly quantified. Existing studies report a wide range of CRB influence (7%-78%) because satellites miss many short-lived or low-intensity fires, chemical signatures of CRB overlap with those from residential biomass burning, and bottom-up inventories lack reliable, region-specific activity data for agricultural burning. These issues highlight the need for a more rigorous, observation-driven framework to isolate and optimise CRB-related emissions robustly. To identify days when receptor-site PM2.5 was influenced by transported emissions from open biomass burning (CRB), satellite fire counts (MODIS and VIIRS) and Lagrangian particle dispersion modelling (FLEXPART) were used. For each day, weighted fire counts (WFC) were calculated by overlapping satellite fire hotspots with FLEXPART back-trajectory sensitivities, assigning greater weight to fires located within high-sensitivity regions of the footprint. The days were then ranked according to their WFC values. Based on this ranking, the study period was partitioned into biomass-burning period (23 October–16 November) and non-biomass-burning period (4–27 September; 7–11 October). PM2.5 emissions for each period were optimised using the analytical inverse modelling system FLEXINVERT, constrained by surface observations and ECLIPSE v6b as the prior emission inventory. The observational constraint was provided by a 32-station monitoring network distributed across Punjab, Haryana, Delhi NCR (National Capital Region), and western Uttar Pradesh states in the northwestern IGP. The results show that during the non-biomass-burning period, posterior emissions showed a ~25% regional increase, ndicating that anthropogenic non-biomass-burning sources—such as brick kilns, agro-processing facilities, small food-processing units, sugar mills, and agricultural energy use—are systematically underestimated in current global emission inventories. During the biomass-burning period, posterior PM2.5 emission fluxes increased up to ~300% relative to the prior. The strongest increments occurred over central Punjab, southern Haryana and the India-Pakistan border region, coinciding with known CRB hotspots. In contrast, Delhi NCR exhibited negative increments, suggesting that prior inventories overestimated emissions over Delhi NCR region while underestimating emissions in upwind agricultural states. By comparing biomass-burning and non-biomass-burning periods, CRB-related emission enhancements in posterior fluxes reached ~250% in several Punjab and Haryana grids, whereas Delhi NCR showed only a ~15% increase. The analysis is further extended to 2023 and 2024 which revealed that although satellite fire counts decreased, posterior optimized fluxes increased, suggesting that satellite fire detections alone underrepresent true CRB activity. The study further quantifies the total amount of crop residue burnt and the effective emission factor per fire to reconcile modelled and observed PM2.5. This work provides the observational constrained regional optimisation of CRB-related PM2.5 emissions over the IGP, offering new insight into the quantification, spatial distribution, and regional influence of CRB emissions in India. These improved CRB emission estimates can provide insight for air-quality mitigation and agricultural-burning policy, and provide an input for aerosol-climate modelling studies.