Improved Subsurface Model Calibration Using Soft Data on Geologic Heterogeneity
Abstract
Calibration of numerical models for groundwater flow and transport uses available field data from observation wells that may be limited in number and located non-optimally. A series of data sets generated in a three- dimensional synthetic aquifer were used to explore the relationship between accuracy of estimated model parameters and increasing quantities of soft data. The synthetic aquifer consisted of five well-characterized sands carefully packed in a tank 208 cm long, 117 cm wide, and 57 cm tall. The sands were packed to represent a stationary spatially correlated random field with a moderate heterogeneity. A lens-like layer of fine sand embedded into the stationary field, resulting in a non-stationary composite aquifer heterogeneity. The presence of a single layer of fine sand was treated as soft data gathered through geophysical characterization. Pressure distribution within the aquifer was measured at 92 monitoring locations using an automated measuring system. Constituent transport data was generated using a new, automated, fluorimetry-based plume monitoring system consisting of 20 channels with optical fibers. The data generated were used to conduct a number studies related to the use of soft and hard data for model calibration. The first study, using the flow simulator MODFLOW and inversion code UCODE, confirmed that the accuracy in parameter estimation improves with increasing quantities of soft data especially information on heterogeneity. This supports the inclusion of a model designed to capture geological complexity to improve the accuracy of parameter estimation. Similar studies are underway and will be compared.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2006
- Bibcode:
- 2006AGUFM.H43A0476H
- Keywords:
-
- 0520 Data analysis: algorithms and implementation;
- 1829 Groundwater hydrology;
- 1831 Groundwater quality;
- 1832 Groundwater transport;
- 1847 Modeling