An empirical model to estimate ammonia emission from cropland fertilization in China

Ammonia (NH3) volatilization is one of the main pathways of nitrogen loss from cropland, resulting not only in economic losses, but also environmental and human health impacts. The magnitude and timing of NH3 emissions from cropland fertilizer application highly depends on agricultural practices, climate and soil factors, which previous studies have typically only considered at coarse spatio-temporal resolution. In this paper, we describe a first highly detailed empirical regression model for ammonia (ERMA) emissions based on 1443 field observations across China. This model is applied at county level by integrating data with unprecedented high spatio-temporal resolution of agricultural practices and climate and soil factors. Results showed that total NH3 emissions from cropland fertilizer application amount to 4.3 Tg NH3 yr(-1) in 2017 with an overall NH3 emission factor of 12%. Agricultural production for vegetables, maize and rice are the three largest emitters. Compared to previous studies, more emission hotspots were found in South China and temporally, emission peaks are estimated to occur three months earlier in the year, while the total amount of emissions is estimated to be close to that calculated by previous studies. A second emission peak is identified in October, most likely related to the fertilization of the second crop in autumn. Incorporating these new findings on NH3 emission patterns will enable a better parametrization of models and hence improve the modelling of air quality and subsequent impacts on ecosystems through reactive N deposition.

Automated summary

This study developed a highly detailed empirical regression model for NH3 emissions based on 1443 field observations in China, showing that total NH3 emissions from cropland fertilizer application amounted to 4.3 Tg NH3 in 2017, with vegetables, maize, and rice being the largest emitters. More emission hotspots were found in South China, and emission peaks were estimated to occur three months earlier than previous studies. Incorporating these findings will improve the modeling of air quality and impacts on ecosystems.