HR: 17:30h
AN: H44A-07 [Abstracts]
TI: Reducing Uncertainty in Coupled Land Surface-Atmosphere Modeling Via Simultaneous State and Parameter
Estimation
AU: * Liu, Y
EM: yqliu@hwr.arizona.edu
AF: Univeristy of Arizona, Dept. Hydrology & Water Resources
845 N Park Avenue, Tucson, AZ 85721
United States
AU: Gupta, H
EM: hoshin.gupta@hwr.arizona.edu
AF: Univeristy of Arizona, Dept. Hydrology & Water Resources
845 N Park Avenue, Tucson, AZ 85721
United States
AU: Anderson, J
EM: jla@ucar.edu
AF: National Center for Atmospheric Research, Mesa Laboratory
1850 Table Mesa Drive, Boulder, CO 80305
United States
AU: Hoar, T
EM: thoar@ucar.edu
AF: National Center for Atmospheric Research, Mesa Laboratory
1850 Table Mesa Drive, Boulder, CO 80305
United States
AB:
Imperfect data and non-effective model parameterizations/parameters are among the major sources of uncertainties in land
surface and atmospheric modeling. A great deal of research has been conducted concerning the uncertainty associated with
model state variables (via data assimilation) or model parameters (via calibration or optimization). However, little or
nothing has been done to examine the combined effects of these two techniques, which we would expect to result in both
improved model states and less biased parameter estimates, thus enhancing land surface modeling to a new level of accuracy
and reliability. In coupled land surface-atmosphere modeling, model-generated precipitation and radiation hold substantial
uncertainty while representing the major driving forces to the underlying land surface processes. Some initial parameter
estimation experiments with a locally coupled land surface-atmosphere model show that both parameter estimation and model
simulations are greatly improved when model-generated precipitation and net radiation are replaced with corresponding
observations at each time step, implying the great potential benefit that can be gained by assimilating observations of these
two variables to optimally extract information from models and data. This study examines the important effects of
assimilating precipitation and radiation observations on simulations and predictions of land surface-atmosphere modeling by
conducting simultaneous state and parameter estimations. The optimization algorithm Shuffled Complex Evolution (SCE) is used
for parameter estimation and the Ensemble Filtering techniques supported by the NCAR Data Assimilation Research Testbed
(DART) are used for state estimation. The model used in this study is the NCAR Single-column Community Climate Model (SCCM);
and data for assimilation and evaluation are from the Southern Great Plains (SGP) site of the Atmospheric Radiation Research
(ARM) program.
DE: 1833 Hydroclimatology
DE: 1840 Hydrometeorology
DE: 1843 Land/atmosphere interactions (1218, 1631, 3322)
DE: 3333 Model calibration (1846)
DE: 3384 Acoustic-gravity waves
SC: Hydrology [H]
MN: Fall Meeting 2005