Reliable detection and attribution of changes in nitrogen (N) runoff from croplands are essential for designing efficient, sustainable N management strategies for future. Despite the recognition that excess N runoff poses a risk of aquatic eutrophication, large-scale, spatially detailed N runoff trends and their drivers remain poorly understood in China. Based on data comprising 535 site-years from 100 sites across China's croplands, we developed a data-driven upscaling model and a new simplified attribution approach to detect and attribute N runoff trends during the period of 1990-2012. Our results show that N runoff has increased by 46% for rice paddy fields and 31% for upland areas since 1990. However, we acknowledge that the upscaling model is subject to large uncertainties (20% and 40% as coefficient of variation of N runoff, respectively). At national scale, increased fertilizer application was identified as the most likely driver of the N runoff trend, while decreased irrigation levels offset to some extent the impact of fertilization increases. In southern China, the increasing trend of upland N runoff can be attributed to the growth in N runoff rates. Our results suggested that increased SOM led to the N runoff rate growth for uplands, but led to a decline for rice paddy fields. In combination, these results imply that improving management approaches for both N fertilizer use and irrigation is urgently required for mitigating agricultural N runoff in China.
Keywords: Attribution analysis; Bayesian inference; Nitrogen runoff; Spatial pattern; Temporal trend.
Copyright © 2017 Elsevier Ltd. All rights reserved.