黄土高原典型流域水沙变化趋势预测集合评估

Recently, affected by the aggravating climate change and human activities, runoff and sediment yield of the Yellow River has decreased significantly. How does the trend of runoff and sediment yield of the Yellow River change in the near future? Does the strategic policy on control of runoff and sediment yield of the Yellow River and Loess Plateau need to be adjusted? Forecasting can be made on the basis of scientific evaluation on the trend of runoff and sediment dynamic of the Yellow River. Based on the existing prediction methods and modeling results on runoff and sediment dynamic in watersheds, this project will carry out a comparative analysis on differences, performance accuracies and acceptances among different prediction theories, input data, indicators, parameters and prediction results. Meanwhile, simulation and modeling of runoff and sediment dynamic by using hydrological method, water conservation method and physical process modeling in typical watersheds at the simultaneous temporal and spatial scales will be analyzed. The uncertainty effects in input data, parameters and model structures are going to be studied by applying Bayesian test, mathematical statistics analysis, etc. The new evaluation technique named the set evaluation technique will be developed to compare different simulation methods and prediction results. The trend of runoff and sediment dynamic with determined the confidence intervals in typical watersheds will be predicted by using the set evaluation technique. The expected studied method could be extent to forecast the trend of runoff and sediment dynamic in the Yellow River basin. Also, this study provides a theoretical basis for making decisions scientifically on the management strategies of Loess Plateau and the Yellow River. The newly set evaluation technique could promote the development of similar multi-model set evaluation techniques in other fields.

受气候变化与人类活动影响加剧，近年黄河水沙锐减。黄河水沙变化未来趋势如何？黄土高原与黄河治理策略是否调整？须在科学评估水沙变化趋势基础上做出评价。本研究基于既有的流域水沙预测方法及已有的预测成果，开展各类方法数学解析与内在机理、输入数据、特征指标、方法结构与预测结果间的差异性、优劣性和认可度对比分析，综合评价各类方法不确定性与其适用性；补充同时空尺度的水文法、水保法、物理过程模型法等水沙变化模拟与预测，采用贝叶斯检验、数理统计等新型理论方法，分析各类方法在输入、参数、结构对输出的不确定性影响，研发可统一比较与集合分析各类方法模拟结果的集合评估技术，开展典型流域水沙变化趋势多维集合评估，集合预报未来水沙变化趋势，确定其置信区间。本研究预期成果可扩展至黄河流域水沙变化趋势预测，为科学评估未来常态水沙情势、科学决策黄土高原与黄河未来治理策略提供重要依据，并将推动其他领域多模型集合评估技术发展。

黄河水沙变化情势深刻影响着黄河流域水沙调控工程布局、流域内外水资源配置和跨流域调水工程建设等黄河保护与治理开发重大问题的决策。但受研究时段、方法及其边界条件等影响，黄河水沙变化预测成果差异大，难形成共识，成为制约黄河水沙未来变化趋势科学研判的重要瓶颈。本项目通过四年攻关研究，在野外观测、数值分析与模拟与集合评估基础上，系统梳理了流域水沙变化归因-预测既有方法，在量化识别各类既有水沙预测方法适用性和预测成果不确定性基础上，创新研发了流域水沙变化趋势预测集合评估技术，并基于多模型方法水沙预测结果，集合评估了黄河流域未来30-50年水沙变化趋势及其置信区间，研究结果可为未来治黄策略制定提供重要的水沙调控边界条件。.本项目全面完成了任务书规定的研究内容和考核指标，达到了预期目标，形成的技术成果如下：（1）既有流域水沙变化归因与预测成果差异大，原因在于不同方法因数据输入、构成变量及精度评价方法差异带来的诸多不确定性；（2）提出了基于输入-结构-输出的模型适用性判别准则和评价技术，并基于标准化的数据输入，从数据需求、物理基础、应用效率、输出尺度和预测精度等五个维度，集合评价了既有水沙变化预测模型的适用性；（3）构建了流域水沙变化多模型集合评价-多结果加权融合-BMA集合预测的集合评估技术，根据9类模型对黄河潼关站沙量预测成果，预测了黄河流域未来30-50年沙量约为3亿t左右。