Abstract
While global sensitivity analysis has been widely used for identifying dominant processes in
geophysical modeling, the existing methods consider only parametric uncertainty but ignore model
uncertainty, which arises when a process can be simulated by multiple conceptual-mathematical
models. To address this problem, this study develops new process sensitivity index by integrating
the model averaging method into the framework of variance-based global sensitivity analysis; the
model averaging method quantifies both parametric and model uncertainty. The new index also
considers interactions between process models and between model parameters. For demonstration,
the new index is used to evaluate the recharge and geology processes of a synthetic study of
groundwater reactive transport modeling. The recharge process is simulated by two alternative
models, and so is the geology process; each process model has its own parameters. The new
process sensitivity index is mathematically general, and can be applied to a wide range of
geophysical problems.Hydrological models are always composed of multiple components that represent processes key to intended model applications. When a process can be simulated by multiple conceptual-mathematical models (process models), model uncertainty in representing the process arises. While global sensitivity analysis methods have been widely used for identifying important processes in hydrologic modeling, the existing methods consider only parametric uncertainty but ignore the model uncertainty for process representation. To address this problem, this study develops a new method to probe multimodel process sensitivity by integrating the model averaging methods into the framework of variance-based global sensitivity analysis, given that the model averaging methods quantify both parametric and model uncertainty. A new process sensitivity index is derived as a metric of relative process importance, and the index includes variance in model outputs caused by uncertainty in both process models and model parameters. For demonstration, the new index is used to evaluate the processes of recharge and geology in a synthetic study of groundwater reactive transport modeling. The recharge process is simulated by two models that converting precipitation to recharge, and the geology process is also simulated by two models of different parameterizations of hydraulic conductivity; each process model has its own random parameters. The new process sensitivity index is mathematically general, and can be applied to a wide range of problems in hydrology and beyond.
