HR: 0800h
AN: H51D-06    [Abstracts]
TI: Why is an LSM augmented with an aquifer model less sensitive to parameter choices than a standard LSM?
AU: * Gulden, L E
EM: gulden@mail.utexas.edu
AF: The Jackson School of Geosciences, Department of Geological Sciences, The University of Texas at Austin 1 University Station #C1100, Austin, TX 78712, United States
AU: Rosero, E
EM: erosero@mail.utexas.edu
AF: The Jackson School of Geosciences, Department of Geological Sciences, The University of Texas at Austin 1 University Station #C1100, Austin, TX 78712, United States
AU: Yang, Z
EM: liang@mail.utexas.edu
AF: The Jackson School of Geosciences, Department of Geological Sciences, The University of Texas at Austin 1 University Station #C1100, Austin, TX 78712, United States
AU: Jackson, C S
EM: charles@utig.ig.utexas.edu
AF: The Jackson School of Geosciences, Institute for Geophysics, J.J. Pickle Research Campus Bldg. 196 10100 Burnet Road (R2200), Austin, TX 78758, United States
AU: Rodell, M
EM: matthew.rodell@nasa.gov
AF: NASA Goddard Space Flight Center, Hydrological Sciences Branch Code 614.3, Greenbelt, MD 20771, United States
AB: Previous work has shown that the addition of a lumped, unconfined aquifer model to a land-surface model (LSM) decreases the sensitivity of modeled monthly change in terrestrial water storage (dTWS) to selection of subsurface hydrologic parameters. In the work presented here, we investigate why adding an aquifer model as the lower boundary condition increases the robustness of LSM-simulated dTWS. We also evaluate the sensitivity of modeled runoff and modeled partitioning of terrestrial water storage to parameter choices. We compare two versions of the National Center for Atmospheric Research Community Land Model (CLM): (1) the standard CLM, which has a 10-layer, 3.43-m soil profile and no explicit aquifer representation; and (2) the standard CLM augmented with a lumped, unconfined aquifer model as its lower boundary. We simulate 1997 to 2005, driving the models offline as a single column representing Illinois, USA. The two versions of CLM are each run 20,000 times using a random sample of the parameter space for soil texture and key hydrologic parameters. We use the model covariance structures to identify how the shift in sensitivity occurs and perform detailed analysis of the interaction between parameters and model processes under a range of hydrologic conditions.
DE: 1631 Land/atmosphere interactions (1218, 1843, 3322)
DE: 1829 Groundwater hydrology
DE: 1833 Hydroclimatology
DE: 1840 Hydrometeorology
DE: 1847 Modeling
SC: Hydrology [H]
MN: 2007 Joint Assembly