AERMOD 대기확산 해석을 통한 축산 밀집 지역의 암모니아 시공간적 해석

The particulate matters (PM10 and PM2.5) and ammonia emitted from livestock farms as dispersed to urban and residential areas can increase the public’s concern over the health problem, social conflicts, and air quality. Understanding the atmospheric dispersion of such matters is important to prevent the problems for the regulatory purposes. In this study, AERMOD modeling was performed to predict the dispersion of livestock particulate matters and ammonia in Gwangju metropolitan city and five surrounding cities. The five cities were divided into 40 sub-zones to model the area-based emissions which varied with the number of livestock farms, species and growth stages of the animals. As a result, the concentrations of PM10, PM2.5 and ammonia resulted from livestock farms located in the surrounding cities were 2.00 µg m-3, 0.30 µg m-3 and 0.04 ppm in the southwestern part of Gwangju based on the average concentration of 1 hour. These values accounted for 0.7% of PM10 concentration, 0.5% of PM2.5 concentration, and 0.4% of the ammonia concentration in Gwangju, contributing to a small amount of air pollution compared to other sources. As preventive measures, the plantation was applied to high emission source areas to reduce particulate matters and ammonia emissions by 35% and 31%, respectively, and resulted in decrease of the area of influence by 57% for particulate matters and 59% for ammonia.

Although the AERMOD model, which is recommended by the U.S. Environmental Protection Agency, is widely used for modeling air dispersion, its use for producing time-series forecast information other than for regulatory purposes is limited, especially for scientific research. The aim of this study is to evaluate the dispersion of livestock-derived NH3 with six different input datasets and validate the AERMOD model based on hourly time-series measurements. The AERMOD model was verified and evaluated using statistical methods such as the index of agreement, the fractional bias, the normalized mean square error, and the fraction of predictions within the factor two observations. Nine out of 12 model predictions were statistically reliable, thus demonstrating the potential of the model for predicting NH3 concentrations from livestock facilities and the usefulness of time-series analysis in this context. The low IA values in some cases indicate a tendency for the AERMOD model to overestimate the predicted values under low wind speeds and stable atmospheric conditions. The AERMOD model was found to perform well in predicting NH3 concentrations at receptors using local meteorological data and the finer town-scale area emission inputs, thereby suggesting its applicability for regulatory purposes as well as time-series predictions. Improving input data and conducting long-term modeling could further improve the performance of the model.