On the assimilation of environmental tracer observations for model-based decision support

Abstract

It has been advocated that history matching numerical models to a diverse range of observation data types, particularly including environmental tracer concentrations and their interpretations and derivatives (e.g., mean age), constitutes an effective and appropriate means to improve model forecast reliability. This study presents two regionalscale modeling case studies that directly and rigorously assess the value of discrete tritium concentration observations and tritium-derived mean residence time (MRT) estimates in two decision-support contexts; "value" is measured herein as both the improvement (or otherwise) in the reliability of forecasts through uncertainty variance reduction and bias minimization as a result of assimilating tritium or tritiumderived MRT observations. The first case study (Heretaunga Plains, New Zealand) utilizes a suite of steady-state and transient flow models and an advection-only particle-tracking model to evaluate the worth of tritium-derived MRT estimates relative to hydraulic potential, spring discharge and river-aquifer exchange flux observations. The worth of MRT observations is quantified in terms of the change in the uncertainty surrounding ecologically sensitive spring discharge forecasts via first-order second-moment (FOSM) analyses. The second case study (Hauraki Plains, New Zealand) employs paired simple-complex transient flow and transport models to evaluate the potential for assimilation-induced bias in simulated surface-water nitrate discharge to an ecologically sensitive estuary system; formal data assimilation of tritium observations is undertaken using an iterative ensemble smoother. The results of these case studies indicate that, for the decision-relevant forecasts considered, tritium observations are of variable benefit and may induce damaging bias in forecasts; these biases are a result of an imperfect model's inability to properly and directly assimilate the rich information content of the tritium observations. The findings of this study challenge the advocacy of the increasing use of tracers, and of diverse data types more generally, whenever environmental model data assimilation is undertaken with imperfect models. This study also highlights the need for improved imperfect-model data assimilation strategies. While these strategies will likely require increased model complexity (including advanced discretization, processes and parameterization) to allow for appropriate assimilation of rich and diverse data types that operate across a range of spatial and temporal scales commensurate with a forecast of management interest, it is critical that increased model complexity does not preclude the application of formal data assimilation and uncertainty quantification techniques due to model instability and excessive run times.

Utilizing Environmental Tracers to Reduce Groundwater Flow and Transport Model Parameter Uncertainties

“…This insight agrees with Knowling et al. (2020) who suggest calibrating against tritium observations can deteriorate model performance when model structural error is large. Model structural errors cause the calibration to compensate through estimation of effective model parameters, which can be differ widely from the modeler's a priori expectation of reasonable parameter values.…”

“…Common ways to numerically simulate groundwater age include particle tracking methods (e.g., LaBolle et al., 1996; Pollock, 2012) or the transport of age mass using the advection‐dispersion equation (Ginn, 1999; Goode, 1996; Varni & Carrera, 1998). Particle tracking ages that only account for advective transport have been compared to groundwater ages that assume piston flow (Portniaguine & Solomon, 1998; Reilly et al., 1994; Sturchio et al., 2014; Szabo et al., 1996) or mean ages inferred using LPM (Gusyev et al., 2014; Knowling et al., 2020). Alternatively, groundwater mean age simulated with random walk particle tracking methods or the advection‐dispersion equation include dispersion‐based mixing processes and are expected to better correspond to mean ages of samples that are characterized by an age distribution (Gardner et al., 2015; Kolbe et al., 2016; Weissmann et al., 2002).…”

Robust Data Worth Analysis with Surrogate Models

“…Many researchers in hydrological sciences have taken on the question of worth of data in recent years (e.g., Cirpka et al 2004; Fu and Gómez‐Hernández 2009; Dausman et al 2010; Wöhling et al 2016; Siade et al 2017; Sreekanth et al 2017; Knowling et al 2020). Tangentially, data worth estimation is often linked to the topic of monitoring network design, which includes the costs of potential measurements as well (e.g., Erickson et al 2002; Feyen and Gorelick 2005).…”

“…Surrogate models is a term that encompasses a multitude of simplification‐methods and stand‐ins, with the common goal to “increase computational efficiency” (Asher et al 2015). Usually, they are divided in three distinct categories (cf Asher et al 2015): projection‐based methods (e.g., Vermeulen et al 2004a; McPhee and Yeh 2008; Pasetto et al 2013; Boyce and Yeh 2014; Ushijima and Yeh 2015; Stanko et al 2016), data‐driven methods (e.g., Khu and Werner 2003; Yoon et al 2011; Taormina et al 2012; Roy et al 2016; Fan et al 2020; Yin and Tsai 2020), and structural simplification methods (e.g., Sun 2008; Doherty and Christensen 2011; Watson et al 2013; von Gunten et al 2014; Knowling et al 2020).…”

Early Uncertainty Quantification for an Improved Decision Support Modeling Workflow: A Streamflow Reliability and Water Quality Example

“…Ultimately, this can compromise the reliability of decisions made on the basis of model forecasts. Additionally, Knowling et al (2020) showed that there is significant potential to induce predictive bias due to the inability of an imperfect model to appropriately assimilate information-rich data, such as environmental tracer observations. Therefore, when faced with using a model with prediction relevant imperfections, a modeler may wish to critically consider whether history matching is appropriate.…”