利用 GLUE 方法不同算法分析 TOPMODEL模型不确定性

以丹江的西峡流域为研究对象，采用GLUE方法探讨TOPMODEL模型不确定性。分析不同似然函数算法、调节参数、模型参数和结果预测区间的不确定性，结果表明不同算法的似然函数值既有差异又有相似性，值变化会引起不同似然函数分布的变化，也影响模型参数与似然值的关系。分析不同阈值的90％置信区间的实测值覆盖率、预测区间宽度及预测区间对称性等特征，结果说明阈值越小，预测区间的可靠性越高，但不确定性范围越大；阈值越大则情况相反。说明选定似然函数和阈值时需综合考虑多方面因素。

Uncertainty Analysis of TOPMODEL Model by Using Different Likelihood Functions Based on the GLUE Method

The GLUE methodology is applied to uncertainty analyses of TOPMODEL , with the Xixia catchment in Danjiang River as the research area .The uncertainties of different likelihood functions , shaping factor , model parameters and prediction interval are analyzed .It is shown that the likelihood values for different functions are different as well as similar , and the different distribution of each different likelihood function the relationship between the parameters and the likelihood measure are affected by a shaping fac -tor .Lastly the prediction interval characteristics of the 90%confidence level at different threshold values are calculated , such as the observation containing ratio , prediction interval width and precition interval symmetry .It is shown that the lower the threshold value is, the higher the reliability of the prediction interval is , and meanwhile the larger while the range of uncertainty is .It is concluded that many factors should be considered in selecting likelihood function and threshold value .