基于贝叶斯理论的SLURP水文模型参数不确定性评估及日径流模拟分析

为探讨贝叶斯理论在水文模型参数不确定性中的应用效果，以三峡库区兴山水文站以上的香溪河流域及其中的两个子流域1991～1995年日径流资料为研究对象，基于马尔科夫蒙特卡洛算法研究了SLURP水文模型参数及水文模拟的不确定性。结果表明，SLURP水文模型中深层地下水的截留常数、最大入渗率、浅层土壤的蓄水容量和深层地下水的初始百分比4个模型参数的不确定性较低；日模拟径流量95%不确定预测区间对实际观测数据的包括率为14.84%，平均相对区间宽度为1.27；径流模拟不确定性主要的贡献组分为壤中流，且子流域水平上的径流模拟不确定性在子流域汇流过程中发生了累积；两个子流域基流模拟的差别不大，其不确定性随时间逐渐降低。

Parameter Uncertainty Analysis and Daily Streamflow Simulation of SLURP Hydrologic Model Based on Bayesian Theory

To evaluate the application effect of Bayesian theory in parameter uncertainty of hydrological modeling, the daily streamflow during 1991 1995 at the Xingshan hydrometric station as well as two sub basins of the Xiangxi River watershed in Three Gorges Reservoir region was simulated through the SLURP model. And the uncertainty analysis was performed by using Markov Chain Monte Carlo method. The obtained posterior distributions of parameters revealed that four parameters have smaller uncertainty. The results showed that 14.84% of the observations were bracketed by 95% uncertain prediction interval, and the average relative interval length was 1.27; the simulation uncertainty with major contributing component of the interflow varied at the sub basin level, and it was found to be accumulated through routing between sub basins; the simulated base flow varied little across sub basin, but its simulation uncertainty decreased with time.