The prediction uncertainty of model simulations caused by hydrologic models comprises several sources of uncertainty. Briefly they are: (1) observed data, (2) data for model calibration and (3) model structure. To identify and recognise the effect which caused by the different sources of uncertainty is extreamly difficult. In this study, the prediction uncertainty which came from the four kinds of uncertainty sources mentioned previously is classified into four categories: (1) system uncertainty, (2) entire uncertainty, (3) inherent uncertainty, and (4) structure uncertainty. The definition and the procedure to recognize and quantify them are described.The methodology started from applying Monte Carlo simulation to add bias item in model input series (rainfall), then rainfall realizations, parameter space, and model outcomes (outflow discharge) under different input uncertainty level are acquired.Instead of sampling the parameter space directly like what Generalized Likelihood Uncertainty Estimation (GLUE) Methodology did, the study here generates the parameter set space by introducing noise item into input data with specified probability distribution. This reflects the truth that parameter uncertainty came from uncertainty of data to hand and the way the model structure responses it.Finally, by examining the interrelationship among model simulation outcomes, model outcomes during calibration and observed watershed response series (discharge), the different categorized uncertainty can be recognized and quantified by a predefined index with its corresponding input uncertainty level.Statistical second moment and an index which originated from Nash coefficient named Model Structure Indicating Index (MSII) is proposed to quantify model structure uncertainty which can be used as a tool for implementing model quantitative comparison (selection). In order to perform a quantitative comparison between hydrologic models, Storage Function Method (SFM) and TOPMODEL are used.The results show that MSII can well reflects the model structure, where a larger value of MSII indicating a poorer structure of hydrologic model. This can be a useful tool for water resources management, hydrological modeling and rainfall runoff simulation over ungauged basin.
抄録全体を表示