Spatial heterogeneity in PM₂.₅ mitigation from a 25% biomass burning scaling over northern Thailand using WRF-Chem

Northern Thailand experiences severe PM₂.₅ pollution during the dry season from biomass burning, yet the response to incremental emission reductions remains poorly quantified. Using WRF-Chem v4.4, we simulated a 25% biomass-burning emission reduction scenario (Red25, i.e., fire emissions scaled to 75% of Base) over northern Thailand during 1–15 February 2020, evaluated against 15 Pollution Control Department (PCD) stations at 07:00 LT within a two-domain nested system (12 km and 4 km; analyzed at 4 km). The Base simulation showed limited quantitative skill (normalized mean bias − 42.9%, station-averaged correlation 0.25) and substantial meteorological biases, including warm (+ 3.58 °C), dry (− 35.23%), and wind-speed (+ 423%) biases, which constrain confidence in the modeled response magnitude. The Red25 scenario revealed pronounced spatial heterogeneity in the modeled, upper-bound decrease: station-level reductions ranged from 92.1% to 95.5% (station-averaged: 25.1 µg m⁻³, 93.8%), largest in high-fire-density provinces and smallest in urban centers. Across the gridded domain, ΔPM₂.₅ ranged from − 0.5 to − 80.6 µg m⁻³ (domain-averaged: −20.5 µg m⁻³, 88.4%). The mitigation signal showed a suggestive, though not statistically significant, association with temperature bias ( r = − 0.40, p = 0.22) and humidity bias ( r = − 0.46, p = 0.15) at 11 co-located stations. This heterogeneity indicates that uniform emission targets would likely yield heterogeneous benefits, and spatially differentiated strategies may be more effective for policy. However, given the model’s limited skill and exploratory design, the simulated reduction magnitude should be interpreted as an upper-bound diagnostic of model sensitivity, not a validated prediction of real-world air-quality improvement.