Forecast First: An Argument for Groundwater Modeling in Reverse

J. White

DOI: https://doi.org/10.1111/gwat.12558

2017-07-03

Ground Water

Abstract:Introduction Numerical groundwater models are important components of groundwater analyses that are used for making critical decisions related to the management of groundwater resources. In this support role, models are often constructed to serve a specific purpose that is to provide insights, through simulation, related to a specific function of a complex aquifer system that cannot be observed directly (Anderson et al. 2015). For any given modeling analysis, several model input datasets must be prepared. Herein, the datasets required to simulate the historical conditions are referred to as the calibration model, and the datasets required to simulate the model’s purpose are referred to as the forecast model. Future groundwater conditions or other unobserved aspects of the groundwater system may be simulated by the forecast model—the outputs of interest from the forecast model represent the purpose of the modeling analysis. Unfortunately, the forecast model, needed to simulate the purpose of the modeling analysis, is seemingly an afterthought—calibration is where the majority of time and effort are expended and calibration is usually completed before the forecast model is even constructed. Herein, I am proposing a new groundwater modeling workflow, referred to as the “forecast first” workflow, where the forecast model is constructed at an earlier stage in the modeling analysis and the outputs of interest from the forecast model are evaluated during subsequent tasks in the workflow.