渭河关中地区农田蒸散发耗水模拟及其变化归因研究

The assessment of regional evapotranspiration (ET) and its variation attribution under changing environment are currently international frontier on water sciences. The planting structure and water-use patterns of farmland have changed a lot in recent decades in most river basins of China under double impacts of the climate change and human activities,. Existing spatial and temporal distributions of farmland ET consumption also take on new patterns accordingly and there is short of a series of mature attribution analysis theory and methods for farmland ET change. Thus, the Guanzhong area of Weihe River basin under the strong influence of climate change and human activities is selected as the study area, a distributed hydrological model which considers the impact of the climate change and human activities, and the two-layer remote sensing ET model are built for daily ET estimation. Then, the Ensemble Kalman Filter (EnKF) will be used to assimilate the both models to construct a data assimilation system for simulate the ET process with high accuracy, and the observed ET data will be used to test and optimize the simulation. The factor of human activity is fully considered to improve the classical climate elastic coefficient method, and the new attribution approach will be used to quantitatively analyze the impact of climate change and human activities on farmland ET, and the uncertainty of these approaches also will be discussed. The study will reveal the mechanism of the temporal and spatial change for farmland water consumption and the mutual impact mechanism with climate change, and will provide a scientific support for efficient utility of farmland water and regional regulation for sustainable water resources.

变化环境下的区域蒸散发估算及其变化归因研究是当今水文科学领域的国际前沿。在气候变化和人类活动双重影响下，中国大部分流域的农田种植结构和用水方式发生了较大变化，农田蒸散发耗水时空演变也呈现新的规律，目前还缺乏成熟的区域蒸散发耗水变化归因分析理论和方法体系。基于此，本项目拟选择渭河流域关中农业耕作区，构建适合该区域的二层遥感蒸散发模型和考虑自然-人文复合过程的分布式水文模型；并以遥感蒸散发模型为观测算子、以水文模型为模型算子构建蒸散发数据同化系统，结合地面实测数据对同化系统验证、优化，获得较高精度的区域农田蒸散发耗水过程；同时，考虑人类活动因子改进经典的气候弹性系数法，定量归因分析气候变化和人类活动对农田蒸散发耗水的影响；通过探讨上述区域农田蒸散发耗水模拟及其变化归因方法的不确定性，实现揭示农田耗水时空演变规律以及受气候变化影响的机制，为农业水资源高效利用和区域水资源可持续管理提供科学支撑。

蒸散发的准确估算和连续获取是水文科学领域的研究难点，更是优化农业灌溉的有效依据，渭河流域位于干旱半干旱地区，流域内关中地区分布着九大灌区，承担着陕西省近半数的粮食生产，灌溉用水的严重缺乏制约了区域经济发展，因此，科学认识区域蒸散发耗水规律以提高农田用水效率是解决当地资源性缺水的有效途径。. 本项目以渭河流域为研究区，利用集合卡尔曼滤波算法，构建以MOD16为观测算子、以分布式水文模型为模型算子的农田蒸散发数据同化系统，结合流域水文观测资料及全球陆面蒸散发估算数据集(MTE)对同化系统进行验证，获得较高时空精度的区域农田蒸散发过程。首先，经检验MOD16数据的精度满足要求，其与水量平衡要素ET的相关系数为0.72，均方根误差为54 mm/a；与MTE数据相关系数0.92，均方根误差14.81 mm/a，可以用作蒸散发数据同化系统的观测算子；其次，结合渭河流域地形、气象水文以及土地利用等特点构建流域分布式水文模型（WHDM），通过与实测流量验证，月过程效率系数、相关系数、水量平衡系数分别为0.768、0.895和1.030，模拟效果较好，作为同化系统的模型算子；最后，借助集合卡尔曼滤波同化算法，利用水文模型中的ET与土壤湿度的时间响应关系模块，建立蒸散发数据同化系统，模拟出时空连续的流域蒸散发过程。同化以后渭河流域ET日过程得到显著优化，决定系数R2由0.27增加到0.81，均方根误差由0.66降至0.30，月过程和年过程模拟精度也有所提高；典型子流域的ET日过程的决定系数R2由0.09增加至0.75，均方根误差由0.88降至0.37，年过程平均相对误差由13%降为9.6%；同时，同化结果修正了ET的空间分布，子流域蒸散发的数值分层明显，改善了空间变异性，一定程度上反映了人类活动对农田蒸散发耗水的影响，提高了流域蒸散发模拟的精度。