A Hierarchical and Patch Dynamics Paradigm for Modeling Complex, 3D Groundwater Systems Across Multiple Scales

0523107Li这项工作将通过增加一个基于PEST的多尺度逆向建模模块来补充国家科学基金会奖0430987。通过该奖项开发的新的“正向”层级补丁动态模拟能力将得到补充，增加一个重要的多尺度反向模拟部分，这将大大加强我们查明和描述复杂的地下系统和地表水界面通量的能力，从而促进系统地利用来自不同来源的多个尺度的昂贵的实地信息。分层逆环境将基于流行的逆建模引擎PEST--一个通用的、与模型无关的非线性参数估计程序。PEST程序将被改编并无缝嵌入到分层建模环境中。它将与整个模型层次结构动态通信，从而消除离线处理、I/O和文件存储的需要。分层逆向环境将允许在任何尺度组合的一个或多个斑块内对流动和/或传输过程进行综合、同时校准。为了校准复杂的系统，用户将以交互和图形的方式告知PEST哪些参数/输入值将在什么尺度上被调整，以及依赖于补丁和尺度的校准目标(例如，头部、通量、浓度等)。然后，PEST将控制分层建模系统，在调整参数值的同时尽可能多次地运行该系统，直到模型输出与相应的现场或实验室测量之间的偏差在加权最小二乘意义上尽可能小为止。该程序将提供一个分层环境，以便以图形方式显示最佳参数值、与最佳参数值相关的不确定性估计、最佳拟合模型结果、模型与测量的残差以及与最佳参数和残差集相关的一套统计数据。

0523107LiThis work would complement NSF Award 0430987 by adding a PEST-Based Module for Multi-Scale Inverse Modeling. The new "forward" hierarchical patch dynamic modeling capability developed through that award would by complemented by adding an important multi-scale inverse modeling component that would significantly enhance our ability to identify and characterize complex subsurface systems and fluxes at surface water interfaces so as to facilitate systematic use of expensive field information across multiple scales from disparate sources. The hierarchical inverse environment will be based on the popular inverse modeling engine - PEST - a general, model independent nonlinear parameter estimation program. The PEST program will be adapted and seamlessly embedded in the hierarchical modeling environment. It will communicate with the entire model hierarchy dynamically, obviating the need for offline processing, I/O, and file storage. The hierarchical inverse environment will allow integrated, simultaneous calibration of flow and/or transport processes within one or more patches across any combination of scales. To calibrate a complex system, a user will inform PEST interactively and graphically which parameters/input values at what scales will be adjusted and the patch and scale dependent calibration targets (e.g., heads, fluxes, concentration, etc.). PEST will then take control of the hierarchical modeling system, running it as many times as it needs to while adjusting parameter values until the discrepancies between model outputs and corresponding field or laboratory measurements are as small as possible in the weighted least squares sense. The program will provide a hierarchical environment to graphically display optimal parameter values, an estimate of the uncertainty associated with optimal parameter values, best-fit model outcomes, model-to-measurement residuals, and a suite of statistics related to the optimal parameter and residual sets.