基于非一致性的多维降尺度模型及其在径流响应中的应用研究

Statistical downscaling is the indispensable process to bridge the gap between the climate model outputs and the hydrological model inputs, which is essential to assessing the climate change impacts on runoff. The reliability of downscaling method directly influences on the accuracy of hydrological projections. However, the current statistical downscaling methods do not take into account the bias nonstationarity of climate model outputs, and the bias in inter-variable and inter-station correlations. Thus, it is difficult to study the temporal and spatial variability of the runoff impacts in a changing climate. Accordingly, taking Hanjiang River watershed as a case study , this project will (1) propose a pseudo observation approach to test the bias nonstationarity of climate model outputs and identify its attribution; (2) propose an approach to identify the degree of bias nonstationarity of climate model outputs and create a multi-dimensional downscaling model with consideration of the bias nonstationarity to generate multi-variable and multi-site climate change scenarios; (3) combine multi-dimensional climate change scenarios and hydrological models of the Hanjiang River watershed to assess the response of runoff to climate change for the Hanjiang River watershed. The outcomes of this project is theoretical significant to increase the reliability of runoff projections in a changing climate. They also have important practical application values for future water resource optimal management and sustainable utilization, as well as the water supply security of the Middle Route Project of the South-to-North Water Diversion Project for the Hanjiang River watershed.

统计降尺度方法是连接气候模式输出与水文模型输入以评估流域径流对气候变化响应的重要纽带，其精度直接影响径流预测的可靠性。但现有的统计降尺度方法未考虑气候模式输出变量偏差的非一致性，以及变量间和站点间的相关性，难以准确揭示气候变化下流域径流响应的时空变异性。本项目将以汉江流域为研究对象，提出基于虚拟观测方法的气候模式输出变量偏差非一致性检验方法，以识别偏差非一致性的成因；提出识别气候模式输出变量偏差非一致性程度的方法，并基于多维概率密度函数转化方法，创建考虑偏差非一致性的多维降尺度模型，以生成流域尺度多变量多站点的未来气候变化情景；结合多维气候变化情景与汉江流域水文模型，评估未来气候变化下汉江流域的径流响应。其研究成果不仅对提高气候变化下流域径流预测的可靠性具有重要的理论意义，而且对于汉江流域未来水资源优化管理与可持续利用以及南水北调中线工程的供水安全等具有重要的实际应用价值。

国家自然科学基金面上项目“基于非一致性的多维降尺度模型及其在径流响应中的应用研究”，以提高气候变化下流域径流预估的可靠性，降低径流预估的不确定性为总体目标。主要研究内容包括：气候模式输出变量的偏差非一致性及其成因；气候模式输出多维变量偏差的非一致性程度；考虑偏差非一致性的多维降尺度模型；基于非一致性多维降尺度模型的汉江流域径流响应。项目严格按照研究计划开展工作，顺利并超额完成了任务计划。基于该项目出版专著1部；发表学术论文49篇，其中SCI论文44篇；培养博士生5人，硕士生6人；参加国内外学术会议30余人次；研究成果支撑获得湖北省科技进步一等奖1项（5/10）。主要研究成果如下：基于虚拟观测的方法检验并识别了气候模式输出变量偏差的非一致性及其成因；通过建立气候模式输出变量的偏差与模拟值之间的函数关系识别了偏差的非一致性程度，并基于分位数匹配和置乱算法，创建了考虑偏差非一致性的多维降尺度模型；揭示了气候变化下汉江流域的径流响应不确定性的变化规律。在理论方法创新和应用方面均取得了较大进展。通过该项目的研究提高了气候变化水文响应预估的可靠性，并降低了不确定性，可为气候变化下流域水资源管理提供科学参考。