Refining ammonia inventories through top-down inverse modelling in high-density swine farming regions

The accurate estimation of ammonia (NH 3 ) emissions from livestock facilities is required to establish effective air quality management policies, yet significant uncertainty remains in current bottom-up inventories. This study thus presents a top-down inverse modelling approach for the estimation of NH 3 emission factors. We validated the performance of three atmospheric dispersion models—AERMOD, CALPUFF, and computational fluid dynamics (CFD)—against atmospheric NH 3 datasets collected over four seasons in a high-density pig farming region with aging, older-type facilities in South Korea. Evaluation of model performance revealed that the CFD model produced the most accurate local concentration distributions, with an index of agreement (IA) of 0.98, while CALPUFF effectively captured long-term temporal variation (IA = 0.89). Based on the CALPUFF results, the average NH 3 emission factor was estimated to be 4.17 \(\:±\:\) 1.33 kg animal − 1 year − 1 , with seasonal fluctuations ranging from 3.50 kg animal − 1 year − 1 in autumn to 4.53 kg animal − 1 year − 1 in winter. The estimated emission factor was around 80% of the legacy Clean Air Policy Support System value, falling between the modernised (2020) and legacy (2008) inventory factors, suggesting the potential of the top-down inverse approach as a complementary tool for refining bottom-up emission inventories. The proposed method can help improve the representativeness of emission estimates by incorporating receptor-based atmospheric observations from large-scale source clusters.