Submitted to: Meeting Abstract
Publication Type: Abstract only
Publication Acceptance Date: 9/1/2012
Publication Date: 12/2/2012
Citation: Pachepsky, Y.A., Guber, A., Gish, T.J., Yakirevich, A., Nicholson, T., Cady, R. 2012. Applying model abstraction techniques to optimize monitoring networks for detecting subsurface contaminant transport. [abstract]. Interpretive Summary:
Technical Abstract: Improving strategies for monitoring subsurface contaminant transport includes performance comparison of competing models, developed independently or obtained via model abstraction. Model comparison and parameter discrimination involve specific performance indicators selected to better understand subsurface contaminant transport to optimize groundwater monitoring networks (GMN). Three abstraction techniques were validated for GMN design: (1) using pedotransfer functions, (2) profile aggregation, and (3) limiting the input domain by ignoring the unsaturated zone. Data were collected in the tracer experiment at the USDA-ARS OPE3 integrated research site. A pulse of a potassium chloride solution was applied to a 13m x14 m irrigation plot, and chloride concentrations were measured in the groundwater at three sampling depths in 12 observations wells installed at distances of 7 m and 14 m from the irrigation plot. The spatial distribution of soil materials was obtained from cores taken at 0.2 m increments to the depth of 2 m during installation of the observation wells. Soil hydraulic conductivity values were obtained from the HYDRUS-3D calibration with chloride concentration time series measured in the observation wells, and soil water retention was estimated from pedotransfer functions. The model abstraction techniques were evaluated using HYDRUS-3D simulations performed for different hydrologic scenarios. These scenarios included three weather, two ground-water depth, and two groundwater slope scenarios, as well as two different locations of the contaminant release selected within the irrigation plot. The weather scenarios were based on 25%, 50% and 75% of the 10-year probability of mean annual precipitation. The monitoring locations for GMN were selected based on three performance indicators: the peak concentration (Cpeak), the time to the peak concentration (Tpeak) and total chemical flux (QC). The monitoring locations were selected based on (a) more frequent, and (b) more probable and persistent appearance of maximum or minimum values of the above performance indicators. Cpeak and QC appeared to be more reliable performance indicators compared to Tpeak. The profile aggregation method was found to be the only abstraction technique that generated a GMN differed from the network obtained using the calibrated HYDRUS-3D model based on Cpeak and QC performance indicators. The outcome of this study provides reasonable assurance that model abstraction techniques can be used to optimize monitoring network strategies, and can provide specific the information for the future data collection and abstraction efforts to optimize a GMN.