模型多参数灵敏度与不确定性分析
收稿日期: 2009-01-03
修回日期: 2009-06-14
网络出版日期: 2010-02-20
基金资助
国家自然科学基金重点项目 (40730632);中国科学院院长奖获得者科研启动专项资金
A multi-parameter sensitivity and uncertainty analysis method to evaluate relative importance of parameters and model performance
Received date: 2009-01-03
Revised date: 2009-06-14
Online published: 2010-02-20
以潮白河为研究区域,探讨了与模型参数及模型模拟性能有关的多参数灵敏度及不确定性分析方法(Multi-Parameter Sensitivity and Uncertainty Analysis, MPSUA)。基于Monte Carlo模拟的多参数灵敏度分析,可以评价模型中多个参数的相对重要性。GLUE不确定性分析则能对模型性能进行量化评估。实例研究表明,通过MPSUA方法,可以减少优化参数的个数。而且,在没有对模型进行参数优化之前,基于MPSUA就可以确定模型的模拟精度。例如同样的模型在潮河可以获得比在白河更高的模拟精度。这种同一模型在不同流域所体现的差异性,一方面是源于模型结构本身的不完善,另一方面则与用于建模的数据误差有关。SCE-UA参数优化结果与MPSUA结果几乎一致,说明本文的参数灵敏度与模型总体性能评估方法比较合理。
王纲胜,夏军,陈军锋 . 模型多参数灵敏度与不确定性分析[J]. 地理研究, 2010 , 29(2) : 263 -270 . DOI: 10.11821/yj2010020009
A Multi-Parameter Sensitivity and Uncertainty Analysis (MPSUA) method is developed to evaluate the relative importance of parameters and model performance. The idea of MPSUA is to couple the Generalized Likelihood Uncertainty Estimation (GLUE) with the Multi-Parameter Sensitivity Analysis (MPSA) based on Monte Carlo simulation. The implementation of MPSA includes the following steps:(1) Running the model using randomly generated parameter sets; (2) Computing the objective function values, which are defined as the sum of squared errors between 'observed' and simulated values. The observed values are the output from model simulations using the median of the characteristic range for each parameter. (3) Identifying the 'acceptable' and 'unacceptable' parameter sets by comparing the objective function values to a given criterion, e.g., the 50% division of all sorted objective functions. The objective function value which is less than the criterion is classified as 'acceptable', otherwise it is classified as 'unacceptable'. (4) Measuring the separating degree between two cumulative distribution curves for "acceptable" and "unacceptable" parameters. Larger discrepancy means higher sensitivity. The case study in the Chaobai River Basin of North China showed that the model performance can be evaluated based on MPSUA, even though the optimum parameter values were unknown. For example, the same model could reach a higher modeling precision for the Chaohe River Basin than that for the Baihe River Basin. Such a difference in model performance is likely caused by both the uncertainty from model structure and the uncertainty from input data. The consistency between parameter optimization by SCE-UA algorithm and MPSUA also illustrated the rationality of the methodology applied in this paper. Further studies can take into account multiple objectives into the MSPUA.
[1] Gupta Hoshin V, Sorooshian Soroosh, Hogue Terri S,et al.Advances in automatic calibration of watershed Models. In: Duan Q.et al. (Eds), Calibration of Watershed Models, Water Sci. and Appl.6, AGU, Washington, DC, 2003, 9~28.
[2] Schultz G A, Engman E T.(Eds.) Remote Sensing in Hydrology and Water Management.Springer Verlag Berlin Heidelberg, Germany, 2000.
[3] Rozos Evangelos, Efstratiadis Andreas, Nalbantis Ioannis,et al. Calibration of a semi-distributed model for conjunctive simulation of surface and groundwater flows. Hydrological Sciences Journal, 2004, 49(5): 819~842.
[4] Duan Q Y.Global optimization for watershed model calibration. In: Duan Q. et al. (Eds), Calibration of Watershed Models, Water Sci. and Appl. 6, AGU, Washington, DC, 2003: 89~102.
[5] Duan Q, Sorooshian S,Gupta V K. Effective and Efficient Global Optimization for Conceptual Rainfall-Runoff Models. Water Resour. Res. , 1992, 28(4): 1015~1031.
[6] Duan Q, Sorooshian S,Gupta V K. Optimal Use of the SCE-UA Global Optimization Method for Calibrating Watershed Models. J. of Hydrol., 1994, 158: 265~284.
[7] Duan Q, Gupta V K,Sorooshian S. A Shuffled Complex Evolution Approach for Effective and Efficient Global Optimization. J. Optim. Theo. and Its Appl. , 1993, 76(3): 501~521.
[8] McCuen Richard H. Modeling hydrologic change: statistical methods. Lewis publishers, 2003, 333~365.
[9] White K L, Chaubey I. Sensitivity analysis, calibration, and validations for a multisite and multivariable SWAT model. Journal of the American Water Resources Association, 2005, 41:1077~1089.
[10] Lenhart T, Eckhardt K, Fohrer N,et al. Comparison of two different approaches of sensitivity analysis. Physics and Chemistry of the Earth, 2002, 27:645~654.
[11] Choi J, Harvey J W, Conklin M. Use of Multi-parameter sensitivity analysis to determine relative importance of factors influencing natural attenuation of mining contaminants. In: Proceedings of the Toxic Substances Hydrology Program Meeting, Charleston, SC, (D. Morganwalp, ed.). 1999, 185~192.
[12] Thorsen M, Refsgaard J C, Hansen S, et al. Assessment of uncertainty in simulation of nitrate leaching to aquifers at catchment scale. Journal of Hydrology,2001, 242: 210~227.
[13] Ajami N K, Duan Q Y, Sorooshian S. An integrated hydrologic Bayesian multimodel combination framework: Confronting input, parameter, and model structural uncertainty in hydrologic prediction. Water Resources Research, 2007, 43(1), doi: 10.1029/2005wr004745.
[14] Feyen L, Vrugt J A, Nuallain B O, et al. Parameter optimisation and uncertainty assessment for large-scale streamflow simulation with the LISFLOOD model. Journal of Hydrology, 2007, 332:276~289.
[15] Beven Keith, Freer Jim. Equifinality, data assimilation, and uncertainty estimation in mechanistic modeling of complex environmental systems using the GLUE methodology. Journal of Hydrology,2001, 249: 11~29.
[16] Demarty J, OttléC, Braud I, et al.Contraining a physically based Soil-Vegetation-Atmosphere Transfer model with surface water content and thermal infrared brightness temperature measurements using a multiobjective approach, Water Resour. Res., 2005, 41, W01011, doi: 10.1029/2004WR003695.
[17] 王纲胜,夏军,牛存稳.分布式水文模拟汇流方法及应用. 地理研究, 2004, 23(2): 175~182.
/
| 〈 |
|
〉 |