驱动数据对流域水文模拟中不同结果要素的影响

The impacts of input data on different simulated results in hydrological modelling

流域水循环中各要素之间高度的非线性关系,使得驱动数据对模拟结果的影响研究变得复杂。结合站点实测和遥感的降水、温度和潜在蒸散发驱动数据,构建了叶尔羌河流域的8个MIKE SHE模型;根据方差分析模型(ANOVA)对模型"输入"与"输出"之间显著性影响检验的结果,以积融雪过程为例,剖析了驱动数据影响的具体表现。结果表明:相比站点数据,TRMM降水驱动的模型导致积雪输出的空间分布差异性更加明显;MODIS温度数据驱动的模型导致了中低山区的永久性薄层积雪覆盖更广,而高山区的储雪量更少;遥感蒸散发驱动的模型在积雪模拟方面并无显著差异。ANOVA对水文系统中要素之间的显著性假设检验,可为明晰驱动数据对水文过程的具体影响方式提供重要前提。

The high nonlinear relationships among the factors in hydrological processes make it difficult to study the effects of input data on the simulated results.Based on the observed and remote sensing precipitation,temperature and potential evapotranspiration(PET),eight MIKE SHE models of the Yarkant River Basin were set up.On the premise of statistical hypothesis testing implemented by Analysis of Variables model(ANOVA),different significant effects of input data on simulated outputs were specified.Furthermore,with the snow simulation as a case,the manifestations of input data’s impacts were analyzed.Results shows that:compared with model forcing by interpolated station data,the spatial deviations of snow storage were more obvious in the TRMM driving model.In the LST driving model,there was a larger snow coverage in the low-middle mountain region and a lower one in the high mountain region.However,little deviations were found between models forcing by observed and remote sensing PET.Also,the uncertainty of input data concealed in insensitive components can be clarified based on ANOVA’s hypotheses testing.