Demonstration of a value of information metric to assess the use of geophysical data for a groundwater application

Abstract

Effective groundwater management requires that decision makers choose strategies for the allocation and conservation of water resources that satisfy the objectives of, and draw support from, multiple stakeholders with complex and often contradictory value judgments. We demonstrate a value of information (VOI) approach to assess the benefits of acquiring geophysical data as part of a groundwater management strategy in light of data uncertainty. VOI is a method for determining if the acquisition of information improves a decision maker’s ability to select an optimal action given value judgments, risk tolerance, and anticipated consequences of the action. As a practical example we examine the uncertainty associated with time-domain electromagnetic (TDEM) data and evaluate the contribution of these data to management of desalination operations in a coastal aquifer where there is a risk of contaminating freshwater resources. The reliability of TDEM data is evaluated using a forward modeling approach to calculate data sensitivity to parameters of interest in the decision analysis. We evaluate the value added by acquiring uncertain data using a standard VOI approach. The analysis shows additional geophysical information could improve groundwater managers’ ability to make decisions about desalination operations and quantifies the contribution of geophysical data to the management decision. We note several measures that can be taken to improve estimates of the data reliability as well as the valuation of alternative actions.

Multiscale geophysical imaging of the critical zone

“…To justify new applications and build confidence in the ability of geophysics to detect the parameters of interest, quantifiable “value of information” is required. Progress has been made in recent years at attempting to quantify the value of geophysical measurements for hydrologic science [e.g., Nenna and Knight , ], but the work has been limited.…”

“…To justify new applications and build confidence in the ability of geophysics to detect the parameters of interest, quantifiable "value of information" is required. Progress has been made in recent years at attempting to quantify the value of geophysical measurements for hydrologic science [e.g., Nenna and Knight, 2013], but the work has been limited. A common disconnect between geophysical models and end users of these data is that images of geophysical property distributions do not directly address CZ science questions, for example, What is the depth of weathering?…”

Comparing Well and Geophysical Data for Temperature Monitoring Within a Bayesian Experimental Design Framework

“…These notions are sequentially conditional, such that, in order to make an economic decision, the information must first lead to a material decision, which depends on the information gained. The idea of VOI was applied and developed in works such as Liu et al (2012) to the remediation of groundwater sites, Trainor-Guitton et al ( 2013) to monitor the detection of CO 2 leaks, Nenna and Knight (2014) to evaluate the benefits of acquiring geophysical data as part of a groundwater management strategy, and a geophysical perspective on VOI is presented by Trainor-Guitton (2014). In addition, a new concept called efficacy of information (EOI), which is similar to VOI but without financial rewards or costs, was recently introduced by Caers et al (2022).…”

Machine Learning for Bayesian Experimental Design in the Subsurface

“…These notions are sequentially conditional, such that, in order to make an economic decision, the information must first lead to a material decision, which depends on the information gained. The idea of VOI was applied and developed in works such as Liu et al (2012) to the remediation of groundwater sites, Trainor-Guitton et al (2013) to monitor the detection of CO 2 leaks, Nenna and Knight (2014) to evaluate the benefits of acquiring geophysical data as part of a groundwater management strategy, and a geophysical perspective on VOI is presented by Trainor-Guitton (2014) As a result, the concept of information is central to this dissertation, and it is necessary to quantify the information contained in the model parameters, model outputs, observed data, and mutual information in order to obtain an objective measure of the amount of information shared by the model and observed data; indeed, quantifying the information is the first step in the VOI framework in order to lead to a practical decision. Furthermore, because the random variables in a Bayesian analysis represent known or unknown information, the concept of information is critical to the application of Bayesian methods to inverse problems.…”