English
引用本文:王福兵,刘红娟,田强龙,等.基于能量不平衡优化PSEB蒸散发模型[J].灌溉排水学报,0,():-.
Wang Fubing,Liu Hongjuan,Tian Qianglong,et al.基于能量不平衡优化PSEB蒸散发模型[J].灌溉排水学报,0,():-.
【打印本页】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
过刊浏览    高级检索
本文已被:浏览 221次   下载 0  
分享到: 微信 更多
基于能量不平衡优化PSEB蒸散发模型
王福兵, 刘红娟, 田强龙, 赵子敬, 邱中齐, 杨泽伟, 魏国孝
兰州大学
摘要:
摘要:【目的】优化PSEB蒸散发模型参数,提高模型性能。【方法】利用兰州大学半干旱区流域地表过程与环境变化野外科学观测站的数据,研究PSEB模型优化问题。采用差分进化自适应算法,其核心思想为贝叶斯理论,在PSEB模型中引入能量不平衡修正项(αRn),通过构造多条马尔科夫链来估计参数的后验信息。运用传统评价指标包括决定系数(R2)、线性回归斜率(Slope)、一致性系数(IA)、模型效率(EF)、平均偏倚误差(MBE)、均方根误差(RMSE),对优化后的PSEB模型性能进行评价。【结果】运用DREAM算法之后,模型的部分参数和能量不平衡修正项(αRn)得到了很好的约束。两种方案在模型校准期线性回归斜率分别为0.76、0.91,均方根误差(RMSE)值分别为91.24、78.33,方案二的斜率相比方案一更接近于1,且方案二的RMSE值相比方案一降低14%,且两种方案的一致性指数(IA)均为0.93。在验证期,两种方案的线性回归斜率分别为0.51、0.54,均方根误差分别为73.14和67.02,相比方案一,方案二的斜率更接近于1,且均方根误差(RMSE)降低了8%,一致性指数(IA)分别为0.87和0.88。【结论】DREAM算法降低了PSEB模型中部分参数和能量不平衡修正项的不确定性,提高了模型性能。
关键词:  蒸散发模型;参数优化;能量不平衡
DOI:
分类号:S271
基金项目:国家自然科学(41471023)

Wang Fubing, Liu Hongjuan, Tian Qianglong, Zhao Zijing, Qiu Zhongqi, Yang Zewei, Wei Guoxiao
Lanzhou University
Abstract:
Abstract: 【Objective】To optimize the parameters of PSEB evapotranspiration model and improve the performance of the model.【Methods】The optimization of PSEB model was studied by using the data of the field scientific observation station of surface process and environmental change in the semi-arid area of Lanzhou University. The differential evolution adaptive algorithm is adopted, and its core idea is Bayesian theory. The correction term of energy imbalance(αRn)is introduced into PSEB model, by constructing multiple Markov chains to estimate the posterior information of parameters. The traditional evaluation indexes including determination coefficient (R2), linear regression slope (Slope), consistency coefficient (IA), model efficiency (EF), mean bias error (MBE) and root mean square error (RMSE) are used to evaluate the performance of the optimized PSEB model.【Results】After using dream algorithm, some parameters and energy imbalance correction items(αRn)of the model is well constrained. During the model calibration period, the linear regression slopes of the two schemes are 0.76 and 0.91 respectively, and the root mean square error (RMSE) values are 91.24 and 78.33 respectively. The slope of scheme 2 is closer to 1 than that of scheme 1, and the RMSE value of scheme 2 is 14% lower than that of scheme 1, and the consistency index (IA) of the two schemes is 0.93. In the validation period, the linear regression slopes of the two schemes are 0.51 and 0.54 respectively, and the root mean square errors are 73.14 and 67.02 respectively. Compared with scheme 1, the slope of scheme 2 is closer to 1, the root mean square error (RMSE) is reduced by 8%, and the consistency index (IA) is 0.87 and 0.88 respectively. 【Conclusion】DREAM algorithm reduces the uncertainty of some parameters and energy imbalance correction items in PSEB model and improves the performance of the model.
Key words:  evapotranspiration model; parameter optimization; energy imbalance