长三角地区农田肥料总氮地表流失率和流失负荷估算

以长江三角洲为研究区域,利用2002年以来公开发表的文献中188组农田总氮(TN)地表径流试验样本,基于Bayesian递归回归树模型建立了长三角地区农田肥料部分(不包括土壤本底含量)总氮流失估算模型.同时,在ArcGIS平台上,估算了2008年长三角地区"三省一市"(上海、江苏、浙江与安徽)1 km×1 km农田肥料总氮地表流失率和流失负荷.结果表明:基于长三角地区化肥施用量、年降雨量、土地利用现状与水系等数据构建的长三角地区农田肥料TN地表流失率估算模型是有效的,模型校准和验证R2分别达到0.820和0.744,模拟结果相对可靠.长三角地区农田TN流失率具有显著的空间分异性,其中位值为3.36%(R50为3.09%~3.63%),主要影响因素为降雨量、土壤有机质含量、土壤粘粒比重、施N量等;相应地,2008年流失负荷为88.1 Gg·a-1(71.9~104.4 Gg·a-1,以N计).TN流失率较高的区域集中在淮河北部及江苏东部沿海区域,流失负荷贡献最大的市依次为盐城、徐州、阜阳、亳州,共占研究区域农田肥料TN流失负荷的41%.

Estimation of total nitrogen runoff rate and loading by agricultural fertilizers in the Yangtze River Delta area

A total of 188 runoff measurements of TN concentration from agricultural fields in the Yangtze River Delta area were reviewedbased on the journal articles published from 2002. A statistical model was developed for estimating TN runoff loss from cropland fertilizer (not including soil background nitrogen) area using Bayesian Recursive Regression Tree approach. Based on this model, TN runoff rate and loading from agricultural fertilizers applicationwere estimated at 1 km×1 km scalein 2008 using ArcGIS software. The preliminary results included that: (i) the coefficients of determination (R²) reached 0.820 and 0.744 for model calibration and verification steps, respectively. Thus, the BRRT-deduced statistical model estimating TN runoff rate from cropland fertilizer by fertilizer application, precipitation, land use, and water system was reliable for spatial simulations; (ii) the TN runoff loss rate from cropland was spatially heterogeneous across the delta, with the median of 3.36% andthe interquartile rangingfrom 3.09% to 3.63%; the factors that significantly influenced TN runoff rate were precipitation, soil organic matter content, soil clay particle content, amount of N fertilizer application, etc;overall, the TN runoff loading in the Yangtze River delta area was 88.1Gg·a^-1(71.9~104.4 Gg·a^-1)in 2008; (iii) the areas with higher TN runoff rate were located along the northern Huai River and eastern Jiangsu Province, while the major contributions of TN runoff loading were Yancheng, Xuzhou, Fuyang and Bozhou, which accounted for 41% of cropland TN runoff loading in total.