Modeling Temporal Variability of Gaseous Ammonia Emissions from Chemical Fertilizer Usage in Midwest USA

摘要

Regional scale modeling of atmospheric particulate matter (PM) formation and reactive nitrogen deposition (Nr) to ecosystems requires knowledge of representative emission inventories of precursor pollutants including ammonia (NH_3). Predictions from these modeling efforts can be improved by improving NH_3 emission inventories at spatial and temporal scales used by chemical transport models (CTMs). Currently, NH_3 emissions from chemical fertilizer usage (CFU) are estimated by combining county-level fertilizer sales with emission factors obtained from the European Environment Agency. The Sparse Matrix Operator Kernel Emissions (SMOKE) model is the tool used by modelers to disaggregate the emissions inventory using spatial surrogates and temporal factors to develop gridded inputs to CTMs. For CFU, these factors are developed using knowledge of cropland location and seasonal crop schedules. A major challenge lies in the use of existing, time averaged temporal factors that should be improved by including the influence of farm practices, meteorological conditions, and soil conditions on NH_3 emissions. Due to scarcity of relevant field measurements, NH_3 emissions under different agricultural management scenarios can be alternatively estimated by using process-based biogeochemical models. In this study, the Denitrification Decomposition model is used to improve existing temporal factors in SMOKE by identifying temporal variations in NH_3 emissions from different chemical fertilizers, in Central Illinois. Our results indicate that daily variations in NH_3 emissions are unique to different fertilizer types, temperature variation and crop-specific schedules. Emissions peak during mid-April around the planting dates of corn and soybeans with a second emission peak in late fall corresponding to planting of winter wheat. More rapid release of NH_3 is observed under usage of anhydrous ammonia as compared to urea, where the later fertilizer results in releases extending between 3-5 days post application. These modeling results are important because they provide inputs to CTMs to improve the prediction of regional atmospheric PM formation and Nr deposition to ecosystems in the Midwestern domain.