输入数据精度与准确性对SWAT模型模拟的影响

以潮河流域为研究区域，利用潮河流域1990~2013年监测数据，构建SWAT模型，在模型结构和参数不改变的情况下，探究输入数据精度（DEM分辨率）与准确性（降水插值）对径流和总氮模拟结果影响.结果显示：DEM分辨率变化（30~300m）对径流及总氮模拟效果不同，对径流模拟影响不明显，纳什系数（ENS）和R²可达到0.87以上；对总氮模拟结果影响较大，分辨率越精细，模拟效果越好.不同水文年，DEM分辨率变化对总氮负荷模拟表现不同.丰水年对总氮负荷影响较大，负荷量差异较为明显，枯水年影响相对较小；不同DEM分辨率下，年均（1993~2002年）总氮负荷强度空间分布相似，高负荷区均位于潮河中游，潮河上游及下游的负荷强度相对较低.不同降水输入数据站点分布、站网密度和准确性不同，流域内降水空间分布差异显著.总体上，基于站点较少的气象站插值数据与雨量站实测降水数据，径流和总氮模拟效果较为接近；基于SWAT官方雨量站插值数据，径流和总氮模拟效果较差.不同降水数据输入情景下，模拟的总氮负荷强度模拟的空间分布差异明显；降水量分布较高的区域，负荷量也较高；不同水文年下，不同降水输入对总氮负荷模拟表现不同.丰水年和枯水年，基于气象站插值数据的总氮模拟结果与基于雨量站实测数据的模拟结果较为接近，而基于SWAT官方雨量站插值的模拟误差较大；平水年，基于SWAT官方雨量站插值的模拟结果较气象站插值数据的模拟结果更好.该研究可为流域开展模型构建提供科学参考和借鉴.

The influence of input data precision and accuracy on SWAT model simulation

The influence of input data precision and accuracy (DEM resolution and precipitation interpolation) on runoff and total nitrogen simulation was studied with runoff and water quality monitoring data of Chaohe watershed from 1990 to 2013 as samples. The results showed that there were different effects of the DEM resolution change (30~300m) on runoff and total nitrogen simulation results. No significant impact on runoff simulation were found, with ENS and R² above 0.87. While greater effects on total nitrogen simulation results were found. The finer the resolution, the better the simulation accuracy. Under different hydrological years, heterogeneous impacts on simulation performance were found on total nitrogen (TN) load with the change of DEM resolution. There were more significant differences in wet years than in dry years. The influence on the spatial distribution of the annual average (1993~2002) total nitrogen load were much less with the change of DEM resolutions. The high load areas were located in the midstream of the Chaohe watershed, and low load areas in the upstream and downstream. There were larger discrepancies in the spatial distribution of precipitation in the watershed under different distribution and density of monitoring sites, and accuracy of precipitation input data. In general, the runoff and total nitrogen simulation results based on the interpolation data of the fewer weather stations were close to results based on monitoring data of rainfall stations, and the simulation performance based on interpolation data of SWAT official rainfall stations was relatively poor. The spatial distribution of simulated total nitrogen intensity load were obviously different with different precipitation data inputs. The higher the precipitation, the higher total nitrogen load intensity. In different hydrological years, the simulation performance of total nitrogen load also differed with different precipitation data inputs. In wet and dry years, simulation results of TN based on the interpolation data of the weather stations were closer to results based on the rainfall station data, while large simulation deviations were found with the official SWAT rainfall station data. In the normal years, the simulation performance based on official SWAT rainfall station data was better than those based on the interpolation data of the weather stations. This research provided a scientific reference for watershed modeling applications regarding input data accuracy and precision.