Uncertainty analysis of groundwater model based on data assimilation北大核心CSCD

黑河流域中下游地下水系统受上游冰冻圈融水和降雨的补给,由气候变暖导致的冰冻圈萎缩致使中下游地下水系统的稳定性面临更多的风险。地下水模型是地下水系统稳定性评估的有效手段,但是地下水模型参数往往存在较大的不确定性。为此,本文提出了基于数据同化算法的不确定性分析方法,通过包含观测资料信息减小模型不确定性。采用所提方法分析了(基于MODFLOW构建)黑河流域中游地下水模型中13个参数的不确定性,讨论了算法超参数的影响及其最优取值,分析了地下水模型参数的不确定性。实验结果证明数据同化算法可有效减小地下水模型参数的不确定性,观测资料的种类与数量对参数不确定性的减小起到重要作用;不同地下水模型参数的不确定性不同,地表水与地下水相互作用频繁的区域参数不确定性较大;含水层渗透系数、含水层给水度以及灌溉回流系数对模型输出的地下水位输出影响显著,河床水力传导系数对模型输出的河流流量影响较大。本研究将为地下水研究提供更加可靠的模型方法,为西北内流区地下水哺育的绿洲生态系统稳定可持续研究提供重要支撑。

Uncertainty analysis of groundwater model based on data assimilation

The groundwater system in the middle and lower reaches of Heihe River basin is replenished by the upper cryosphere meltwater and precipitation. The stability of the groundwater system in the middle and lower reaches of Heihe River basin faces more risks due to the cryosphere shrinkage caused by climate warming. Groundwater models are efficient tools for researchers in accessing the stability of groundwater system. However， the uncertainty of groundwater model parameters is an important issue. In this paper， an uncertainty analysis method based on data assimilation algorithm which reduces parameters uncertainty by including observations is proposed. The proposed method is then used to analyze the uncertainty of 13 parameters in a groundwater model of the middle reaches of Heihe River Basin （which is constructed based on MODFLOW）. The influence and optimal values of hyper-parameters are discussed. The uncertainty of groundwater model parameters is analyzed using the proposed algorithm. The results show that the data assimilation algorithm can effectively reduce the uncertainty of groundwater model parameters. The diversity and quantity of observations play an important role in reducing the uncertainties. The parameters in subzones with frequent interaction between surface water and groundwater are more uncertain. The hydraulic conductivity， specific yield and irrigation backflow coefficient have significant influence on the groundwater level， while streambed hydraulic conductivity shows great influence on the streamflow. This study will provide a more reliable modelling method for groundwater research and an important support for researching the stability of groundwater system in the northwestern area.