Bridging groundwater models and decision support with a Bayesian network

Abstract

[1] Resource managers need to make decisions to plan for future environmental conditions, particularly sea level rise, in the face of substantial uncertainty. Many interacting processes factor in to the decisions they face. Advances in process models and the quantification of uncertainty have made models a valuable tool for this purpose. Long-simulation runtimes and, often, numerical instability make linking process models impractical in many cases. A method for emulating the important connections between model input and forecasts, while propagating uncertainty, has the potential to provide a bridge between complicated numerical process models and the efficiency and stability needed for decision making. We explore this using a Bayesian network (BN) to emulate a groundwater flow model. We expand on previous approaches to validating a BN by calculating forecasting skill using cross validation of a groundwater model of Assateague Island in Virginia and Maryland, USA. This BN emulation was shown to capture the important groundwater-flow characteristics and uncertainty of the groundwater system because of its connection to island morphology and sea level. Forecast power metrics associated with the validation of multiple alternative BN designs guided the selection of an optimal level of BN complexity. Assateague island is an ideal test case for exploring a forecasting tool based on current conditions because the unique hydrogeomorphological variability of the island includes a range of settings indicative of past, current, and future conditions. The resulting BN is a valuable tool for exploring the response of groundwater conditions to sea level rise in decision support.

Decision Support Systems and Processes for Groundwater

“…For example, the groundwater decision support system (GWDSS) presents a hybridized example for water allocation that includes both simulation-optimization and lumped parameter modelling tools Pierce et al 2006). Artificial Neural Networks (ANN), such as the River GeoDSS (Triana et al 2010) and Bayesian networks (Molina et al 2013a, b;Fienen et al 2013) present an advanced area of research that leverages algorithms to generate potential candidate solutions. The first report of an immersive environment is implemented for a case in the Sichuan Province, China demonstrating a framework that links virtual environments with models (Zhang et al 2013).…”

“…The application of Bayesian networks (Moura et al 2011;Molina et al 2013a, b;Fienen et al 2013) across multiple cases demonstrates a replicable methodology, and the WEAP-MODFLOW software tool (Le Page et al 2012;Hadded et al 2013) is gaining traction across several applications.…”

A Bayesian belief network approach for assessing uncertainty in conceptual site models at contaminated sites

“…BBNs have also been applied to a wide range of other problems within the field of hydrology and water management. For example, Chan et al [2010] used a BBN for assisting catchment-based water resources management, Wang et al [2009] developed a BBN for assessing and managing farm irrigation systems, while Fienen et al [2013] used a BBN with a numerical groundwater model to study the response of groundwater to sea level rise.…”

Uncertainty Estimation and Evaluation of Shallow Aquifers’ Exploitability: The Case Study of the Adige Valley Aquifer (Italy)

“…Decision Support Systems (DSS) for groundwater applications have been the focus of several studies assessing aquifers vulnerability to potential or effective contaminations [6][7][8] and the sustainability of actual or planned water extractions [9][10][11][12], taking into account also economic constraints [13,14] and climate change scenarios [15]. The present work focuses on the effects of some stress conditions for the aquifer on three assessment criteria: the commonly used depth to water (DTW) and recharge/discharge criteria, supplemented by a newly introduced sustainability parameter (S).…”

“…These studies have evidenced the importance of including into the model as many constraints as possible and the benefit of using soft data, such as hydrofacies geometry, provided that they do not overwhelm the effect of primary observational data, to improve the conceptual model and limit the range of variability of the parameters [29][30][31][32][33][34][35][36]. Even considering such additional sources of information, model uncertainty may still significantly affect DSS outcomes [7,[37][38][39][40][41][42] and uncertainty analysis cannot be avoided in the implementation of water management policies.…”