Extended Abstract (184K)4.13
Evaluating the ability of the Ensemble Transform Kalman Filter to predict the influence of observations on 3-6 day forecasts
Kathryn J. Sellwood, Univ. of Miami/RSMAS, Miami, FL; and S. J. Majumdar and D. Hodyss
In recent years, targeted observations have been deployed in data-sparse regions in order to reduce errors in model initial conditions that may lead to the degradation of short-range (1-3 day) forecasts. The ability of these observations to influence forecasts in the medium range (3-7 days), and the effectiveness of strategies to select optimal targeting locations on these timescales has not yet been explored. One such strategy, the Ensemble Transform Kalman Filter (ETKF) is evaluated here using dropwindsonde data collected during the 2005 and 2006 winter storm reconnaissance programs. The effectiveness of the ETKF is evaluated based on its ability to predict the variance of forecast "signals" produced by these dropwindsonde data, and the resulting reduction of forecast error variance. Factors that influence the performance of the ETKF include the prevailing flow regime and the size and quality of the ensemble forecasts that are employed by the ETKF. A better understanding of the strengths and weaknesses of the ETKF will allow us to predict when targeted observations might be useful for medium- and perhaps longer-range forecasts, in addition to those instances when the deployment of supplementary observations would be likely to have minimal value.
Session 4, Advanced Methods for Data Assimilation
Tuesday, 16 January 2007, 1:30 PM-5:15 PM, 208
Previous paper
Recorded presentation