Parameter optimization of the QUAL2K model for a multiple-reach river using an influence coefficient algorithm
Parameter optimization of the QUAL2K model for a multiple-reach river using an influence coefficient algorithm
复制标题
DOI：
10.1016/j.scitotenv.2010.01.025
复制
发表时间：
2010-03-15
影响因子：
9.8
通讯作者：
Ha, Sung Ryong
中科院分区：
文献类型：
--
作者：
Cho, Jae Heon;Ha, Sung Ryong
采用影响系数算法和遗传算法（GA）建立了最新版QUAL2E河川水质模型QUAL2K的自动定标模型。采用影响系数算法对非定常明渠流进行了参数优化。GA。用于求解优化问题，非常简单易懂，但仍然适用于任何复杂的数学问题，它可以快速有效地找到全局最优解。使用先前建立的QUAL2Kw模型对QUAL2K进行自动校准。将本研究开发的基于影响系数和遗传算法（POMIG）的参数优化方法和QUAL2Kw分别应用于江陵南大川多河段，并对两种模型的结果进行了比较。在建模中，基于水质和水力特性的考虑，将河段划分为两部分。POMIG的校准结果表明，大部分水质变量的计算值与观测值吻合良好。在POMIG和QUAL2Kw的应用中，对河底溶解氧（DO）和叶绿素a （Chl-a）的观测值与预测值存在较大误差；因此，采用2个权重因子（1和5）对下游DO和Chl-a进行加权。权重因子为5时，DO和Chl-a的误差总和略低于权重因子为1时的误差总和。然而，当权重系数为5时，其他水质变量的误差总和比权重系数为1时略有增加。一般来说，POMIG的结果略好于QUAL2Kw。(C) 2010 Elsevier B.V.版权所有
An influence coefficient algorithm and a genetic algorithm (GA) were introduced to develop an automatic calibration model for QUAL2K, the latest version of the QUAL2E river and stream water-quality model. The influence coefficient algorithm was used for the parameter optimization in unsteady state, open channel flow. The GA. used in solving the optimization problem, is very simple and comprehensible yet still applicable to any complicated mathematical problem, where it can find the global-optimum solution quickly and effectively. The previously established model QUAL2Kw was used for the automatic calibration of the QUAL2K. The parameter-optimization method using the influence coefficient and genetic algorithm (POMIG) developed in this study and QUAL2Kw were each applied to the Gangneung Namdaecheon River, which has multiple reaches, and the results of the two models were compared. In the modeling, the river reach was divided into two parts based on considerations of the water quality and hydraulic characteristics. The calibration results by POMIG showed a good correspondence between the calculated and observed values for most of water-quality variables. In the application of POMIG and QUAL2Kw, relatively large errors were generated between the observed and predicted values in the case of the dissolved oxygen (DO) and chlorophyll-a (Chl-a) in the lowest part of the river; therefore, two weighting factors (1 and 5) were applied for DO and Chl-a in the lower river. The sums of the errors for DO and Chl-a with a weighting factor of 5 were slightly lower compared with the application of a factor of 1. However, with a weighting factor of 5 the sums of errors for other water-quality variables were slightly increased in comparison to the case with a factor of 1. Generally, the results of the POMIG were slightly better than those of the QUAL2Kw. (C) 2010 Elsevier B.V. All rights reserved.