Effect of different transport observations on inverse modeling results: case study of a long-term groundwater tracer test monitored at high resolution

Abstract

Conservative tracer experiments can provide information useful for characterizing various subsurface transport properties. This study examines the effectiveness of three different types of transport observations for sensitivity analysis and parameter estimation of a three-dimensional site-specific groundwater flow and transport model: conservative tracer breakthrough curves (BTCs), first temporal moments of BTCs (m1), and tracer cumulative mass discharge (Md) through control planes combined with hydraulic head observations (h). High-resolution data obtained from a 410-day controlled field experiment at Vandenberg Air Force Base, California (USA), have been used. In this experiment, bromide was injected to create two adjacent plumes monitored at six different transects (perpendicular to groundwater flow) with a total of 162 monitoring wells. A total of 133 different observations of transient hydraulic head, 1,158 of BTC concentration, 23 of first moment, and 36 of mass discharge were used for sensitivity analysis and parameter estimation of nine flow and transport parameters. The importance of each group of transport observations in estimating these parameters was evaluated using sensitivity analysis, and five out of nine parameters were calibrated against these data. Results showed the advantages of using temporal moment of conservative tracer BTCs and mass discharge as observations for inverse modeling.

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

“…Furthermore, we expect the above-mentioned issues associated with imperfect-model data assimilation to be relevant and largely transferrable to the assimilation of other environmental tracers, other information-rich observations and diverse data types more generally. This is because we consider the primary barrier to the appropriate assimilation of tritium observation data encountered in the second case study to be fundamental challenges associated with extracting appropriate information from spatially discrete concentration observations when using upscaled or simplified representations of hydraulic properties within a regional-scale model that simulates tracer concentrations using the advection-dispersion equation (e.g., Zheng and Gorelick, 2003;Riva et al, 2008). To the extent that simulated outputs corresponding to observed tracer concentrations are sensitive to model details or parameters that are "missing" in a simplified model (e.g., White et al, 2014), parameter compensation will occur (e.g., Clark and Vrugt, 2006).…”

“…The navigation of this trade-off is central to effective and efficient decision-support modeling practice. In the meantime, tracerdata model assimilation should involve processing or transforming of concentrations into quantities that may be more useful and may guard against ill effects of history matching imperfect models (e.g., by integrating observations in space and time) (e.g., Rasa et al, 2013;Knowling et al, 2019;White et al, 2020).…”

Beyond Classical Observations in Hydrogeology: The Advantages of Including Exchange Flux, Temperature, Tracer Concentration, Residence Time, and Soil Moisture Observations in Groundwater Model Calibration

“…The benefits and the problems that are associated with these assumptions when evaluating model predictive uncertainty and data worth were discussed and compared to the nonlinear and more robust but computationally much more demanding Markov chain Monte Carlo method by Gallagher and Doherty (). Sensitivity analyses (see Hill & Tiedeman, ; Gannett et al, ; Masbruch et al, ; Rasa et al, ), influence statistics (see Hill & Tiedeman, ; La Vigna et al, ), and singular‐value‐decomposition‐based principal component analysis (see Doherty, ; Doherty & Hunt, ; Schilling et al, ) provide additional computationally efficient means to estimate the information content of specific observations or observation types. All these methods are based on the sensitivity matrices calculated with automated flow model calibration routines that are based on weighted multivariate objective functions.…”

Global sampling to assess the value of diverse observations in conditioning a real‐world groundwater flow and transport model

“…On the other hand, many researchers have suggested using additional observational data to constrain the inverse problem. Among other concentration data [e.g., Barlebo et al, 1998Barlebo et al, , 2004, conservative tracers [e.g., Anderman et al, 1996;Barth et al, 2001;Rasa et al, 2013], temperature [e.g., Bravo et al, 2002;Risley et al, 2010], geophysical data [Beaujean et al, 2014], age tracers [e.g., Ginn et al, 2009;Gusyev et al, 2013; Several studies have quantified the value of different observational data in constraining parameter uncertainty in groundwater models. With nonlinear regression being the dominant parameter estimation method in groundwater modeling, most studies used regression results and associated statistics to report on data value [e.g., Barlebo et al, 1998;Barth and Hill, 2005b;Hunt et al, 2006].…”

“…Results therefore do not necessarily hold for possible equally fit models in different regions of the parameter space, as the shape of the response surface will likely vary [Beven, 2009;Rakovec et al, 2014]. The use of more computationally demanding parameter estimation methods, to more inclusively account for model and measurement uncertainty, is increasingly reported in the literature [e.g., Feyen et al, 2003;Hendricks Franssen et al, 2003;Keating et al, 2010;Rojas et al, 2010;Rasa et al, 2013;Carniato et al, 2014]. However, studies that compare different observational data types using more computationally demanding parameter estimation methods are generally based on synthetic model experiments that necessarily idealize the complex reality and disregard disinformation in the observational data.…”