Assimilation of historical head data to estimate spatial distributions of stream bed and aquifer hydraulic conductivity fields. Issue 7 (15th February 2017)

• Record Type: Journal Article
• Authors: Alzraiee, Ayman H.; Bailey, Ryan; Bau, Domenico
• Is Part Of: Hydrological processes. Volume 31:Issue 7(2017)
• Page Start: 1527
• Page End: 1538
• Publication Date: 2017-02-15
• Subjects: conductivity field inversion -- data assimilation -- stream‐aquifer interaction; Hydrology -- Periodicals; Hydrology -- Research -- Periodicals; Hydrologic models -- Periodicals; Hydrological forecasting -- Periodicals; 631.432
• DOI: 10.1002/hyp.11123
• Languages: English
• ISSNs: 0885-6087

Abstract

Abstract: Management of water resources in alluvial aquifers relies mainly on understanding interactions between hydraulically connected streams and aquifers. Numerical models that simulate this interaction often are used as decision support tools for water resource management. However, the accuracy of numerical predictions relies heavily on unknown system parameters (e.g., streambed conductivity and aquifer hydraulic conductivity), which are spatially heterogeneous and difficult to measure directly. This paper employs an ensemble smoother to invert groundwater level measurements to jointly estimate spatially varying streambed and alluvial aquifer hydraulic conductivity along a 35.6‐km segment of the South Platte River in Northeastern Colorado. The accuracy of the inversion procedure is evaluated using a synthetic experiment and historical groundwater level measurements, with the latter constituting the novelty of this study in the inversion and validation of high‐resolution fields of streambed and aquifer conductivities. Results show that the estimated streambed conductivity field and aquifer conductivity field produce an acceptable agreement between observed and simulated groundwater levels and stream flow rates. The estimated parameter fields are also used to simulate the spatially varying flow exchange between the alluvial aquifer and the stream, which exhibits high spatial variability along the river reach with a maximum average monthly aquifer gain of about 2.3 m 3 /day and a maximum average monthly aquifer loss of 2.8 m 3 /day, per unit area of streambed (m 2 ). These results demonstrate that data assimilation inversion provides a reliable and computationally affordable tool to estimate the spatial variability of streambed and aquifer conductivities at high resolution in real‐world systems.