HR: 1340h
AN: H43A-0496    [Abstracts]
TI: One-Dinemsional compared to Three-Dinemsional Snow Data Assimilation
AU: * Clark, M P
EM: clark@vorticity.colorado.edu
AF: Cooperative Institute for Research in Environmental Sciences (CIRES), Campus Box 488 University of Colorado, Boulder, CO 80309-0488 United States
AU: Slater, A G
EM: aslater@cires.colorado.edu
AF: Cooperative Institute for Research in Environmental Sciences (CIRES), Campus Box 488 University of Colorado, Boulder, CO 80309-0488 United States
AB: Data assimilation offers perhaps the best method of alleviating the problem of initial conditions errors when forecasting snowpack or streamflow. There are numerous methods available for assimilation, each with advantages and disadvantages. In this study we look at the relative merits of a 1-Dimensional ensemble kalman filter (EnKF) compared to that of a 3-Dimensional EnKF for use in a distributed hydrologic or snow model. For points without observations, we use an interpolation method to supply observations to the filter when assimilating in the 1-D case. Under the 3-D case we propagate the available observations to unobserved points via an extended state vector in the EnKF. The interpolation relies on locational and topographic attributes, while information propagation in a 3-D filter relies upon establishing the correct covariant relationship between model points. We test our methods using data from SNOTEL stations in the Upper Colorado River Basin.
DE: 1816 Estimation and forecasting
DE: 1840 Hydrometeorology
DE: 1847 Modeling
DE: 1863 Snow and ice (0736, 0738, 0776, 1827)
SC: Hydrology [H]
MN: Fall Meeting 2005