基于扰动分析方法的AnnAGNPS模型水文水质参数敏感性分析

Sensitivity Analysis of AnnAGNPS Model's Hydrology and Water Quality Parameters Based on the Perturbation Analysis Method

参数敏感性分析对于水文水质综合模型的构建及推广应用具有重要的意义.本研究基于AnnAGNPS模型机制,选取可能对模型产生影响的地形、水文气象、田间管理、土壤等4大类31个参数,以太湖地区典型低山丘陵小流域——中田河流域为实验区,利用扰动分析方法评价了模型参数对水文水质模拟结果的敏感性.结果表明,11个地形参数中,LS对径流、泥沙与氮磷等模拟结果普遍敏感,RMN、RS和RVC对泥沙输出分别为一般敏感和弱敏感,其余参数不敏感.水文气象参数中CN对模型径流和泥沙输出特别敏感,对其余结果也都比较敏感.田间管理与植被作物肥料参数中CCC、CRM和RR对泥沙和颗粒态污染物输出弱敏感,6个肥料参数(FR、FD、FID、FOD、FIP、FOP)对氮磷营养盐输出特别敏感.土壤参数中,K对模型除径流之外的所有结果都特别敏感,土壤氮磷含量的4个参数(SONR、SINR、SOPR、SIPR)对相对应的氮磷输出结果弱敏感.通过对中田河流域2005~2010年径流量的模拟和校正检验发现,模拟期与校正期年径流量偏差基本都在10%以内,模拟精度优秀.本研究对AnnAGNPS模型构建中的参数选取与校验调整具有直接参考价值,对研究区的径流模拟结果也证明了模型敏感性分析对于调参过程的可行性和该模型在太湖地区低山丘陵流域径流模拟的适用性,对于模型在国内的推广应用具有重要的指导意义.

Sensitivity analysis of hydrology and water quality parameters has a great significance for integrated model's construction and application. Based on AnnAGNPS model's mechanism,terrain,hydrology and meteorology,field management,soil and other four major categories of 31 parameters were selected for the sensitivity analysis in Zhongtian river watershed which is a typical small watershed of hilly region in the Taihu Lake,and then used the perturbation method to evaluate the sensitivity of the parameters to the model's simulation results. The results showed that: in the 11 terrain parameters,LS was sensitive to all the model results,RMN,RS and RVC were generally sensitive and less sensitive to the output of sediment but insensitive to the remaining results. For hydrometeorological parameters,CN was more sensitive to runoff and sediment and relatively sensitive for the rest results. In field management,fertilizer and vegetation parameters,CCC,CRM and RR were less sensitive to sediment and particulate pollutants,the six fertilizer parameters( FR,FD,FID,FOD,FIP,FOP) were particularly sensitive for nitrogen and phosphorus nutrients. For soil parameters,K is quite sensitive to all the results except the runoff,the four parameters of the soil's nitrogen and phosphorus ratio( SONR,SINR,SOPR,SIPR) were less sensitive to the corresponding results. The simulation and verification results of runoff in Zhongtian watershed show a good accuracy with the deviation less than 10% during 2005-2010. Research results have a direct reference value on AnnAGNPS model's parameter selection and calibration adjustment. The runoff simulation results of the study area also proved that the sensitivity analysis was practicable to the parameter's adjustment and showed the adaptability to the hydrology simulation in the Taihu Lake basin's hilly region and provide reference for the model's promotion in China.