项目名称: 基于信息熵的分布式物理性流域水文模型不确定性分析
项目编号: No.51309011
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 水利工程
项目作者: 龚伟
作者单位: 北京师范大学
项目金额: 25万元
中文摘要: 水文模型广泛存在不确定性。减小不确定性,提升模型在全球变化条件下的模拟能力,是分布式物理性水文模型的发展方向。然而,目前通过不确定性研究改进模型的构想尚未实现,而且在不确定性的量级和来源等基础性问题上仍存在争议。争议的核心问题是:不确定性究竟是源于观测精度不足,还是模型结构有问题。在关于不确定性的讨论中,"信息"是一个常常被提及却鲜见精确定义的概念。针对这一现状,本研究提出基于"信息熵"给出信息的严格定义,拟开展如下四个方面的研究:(1)计算数据提供的信息量,评估数据提供的信息不足导致的不确定性(随机不确定性);(2)计算表示模型结构所需要的信息量,评估由于模型结构导致的不确定性(认知不确定性);(3)提出基于信息熵的模型结构评价方法;(4)应用上述成果改进分布式水文模型。本项研究的目标是,通过定量评估不确定性的量级与来源,最终得到既符合物理过程机制,又有强大模拟能力的分布式物理性模型。
中文关键词: 水文模型;不确定性量化;信息熵;参数优化;替代模型
英文摘要: Every hydrological model has uncertainty. Reducing uncertainty and improving the simulation ability under global change is a frontier of hydrological science. However, the idea of improving model structure through uncertainty study has not yet been achieved. Moreover, the basic problems about uncertainty magnitude and sources are still controversial. The core of these problems is: does the uncertainty comes mainly from observation error, or the model structure error. In current discussion, 'information' is a frequently referred but not formally defined concept. To solve this problem, this study will give a formal definition of 'information' based on 'Shannon's information entropy' and intends to carry out the following four aspects of research: (1) Qualify the information content of data and evaluate the uncertainty caused by observation error (aleatory uncertainty); (2) Qualify the information amount to express the model structure and evaluate the uncertainty caused by model structure; (3) Propose a model structure scrutiny method based on information entropy; (4) Use the methods above to improve a distributed hydrological model. The objective of this research is qualitatively evaluating the magnitude and sources of uncertainty to obtain a physically based, good performance hydrological model.
英文关键词: hydrological model;uncertianty quantification;information entropy;parameter optimization;surrogate model