首页> 外文OA文献 >Coupled machine learning and the limits of acceptability approach applied in parameter identification for a distributed hydrological model
【2h】

Coupled machine learning and the limits of acceptability approach applied in parameter identification for a distributed hydrological model

机译:耦合机学习与分布水文模型参数识别中的可接受性方法的限制

代理获取
本网站仅为用户提供外文OA文献查询和代理获取服务,本网站没有原文。下单后我们将采用程序或人工为您竭诚获取高质量的原文,但由于OA文献来源多样且变更频繁,仍可能出现获取不到、文献不完整或与标题不符等情况,如果获取不到我们将提供退款服务。请知悉。

摘要

Monte Carlo (MC) methods have been widely used in uncertaintyanalysis and parameter identification for hydrological models. The mainchallenge with these approaches is, however, the prohibitive number of modelruns required to acquire an adequate sample size, which may take from days tomonths – especially when the simulations are run in distributed mode. In thepast, emulators have been used to minimize the computational burden of theMC simulation through direct estimation of the residual-based responsesurfaces. Here, we apply emulators of an MC simulation in parameteridentification for a distributed conceptual hydrological model using twolikelihood measures, i.e. the absolute bias of model predictions (Score) andanother based on the time-relaxed limits of acceptability concept (pLoA).Three machine-learning models (MLMs) were built using model parameter setsand response surfaces with a limited number of model realizations (4000). Thedeveloped MLMs were applied to predict pLoA and Score for a large set ofmodel parameters (95 000). The behavioural parameter sets were identifiedusing a time-relaxed limits of acceptability approach, based on the predictedpLoA values, and applied to estimate the quantile streamflow predictionsweighted by their respective Score. The three MLMs were able to adequatelymimic the response surfaces directly estimated from MC simulations with anR2 value of 0.7 to 0.92. Similarly, the models identified using thecoupled machine-learning (ML) emulators and limits of acceptability approach have performedvery well in reproducing the median streamflow prediction during thecalibration and validation periods, with an average Nash–Sutcliffe efficiency value of 0.89 and 0.83, respectively.
机译:Monte Carlo(MC)方法已被广泛用于水文模型的不确定性和参数鉴定。然而,具有这些方法的主要拍卖是获取足够的样本大小所需的巨大模型数量,这可能需要几天的模拟 - 尤其是在分布式模式下运行时。在斯文格中,仿真器通过直接估计基于残留的响应措施来最小化HIMC仿真的计算负担。在这里,我们在参数识别中应用MC模拟的仿真器,用于使用Twolikelious测量,即模型预测(得分)和基于可接受性概念(PLOA)的时间放松限制的模型预测(得分)的绝对偏差.Three机器学习模型(MLMS)使用Model参数SetSand响应曲面构建,具有有限数量的模型实现(4000)。施工的MLMS被应用于预测PLOA并得分为大型模型参数(95 000)。基于预测的平面值,识别行为参数集的可接受性方法的时间放松限制,并应用于估计由它们各自分数的分位式流流程预测。三个MLMS能够充分模拟从MC模拟直接估计的响应表面,ANR2值为0.7至0.92。类似地,使用ThoupleD机器学习(ML)仿真器和可接受性方法的限制来识别的模型在重复和验证周期期间再现中值流出预测,分别为0.89和0.83的平均NASH-Sutcliffe效率值。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
代理获取

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号