Extended Abstract (532K)JP1.18
Use of adjoint-derived sensitivities in constructing an ensemble of forecasts
Daryl T. Kleist, Univ. of Wisconsin, Madison, WI; and M. C. Morgan and G. A. Postel
As part of a study of adjoint-based forecast sensitivities, the MM5v2 adjoint modeling system is being run in real-time at the Univ. of Wisconsin-Madison. The output of the adjoint model, the forecast sensitivities of a variety of forecast aspects to model initial conditions, is posted twice daily on the web site http://helios.aos.wisc.edu. A goal of this study is to understand the relationship between forecast sensitivities and particular characteristics of the synoptic flow pattern from which the sensitivity calculations are performed. In addition, this study seeks to determine the efficacy in using forecast sensitivities in the creation of short-range ensemble forecast products.
In this presentation, an ensemble of short-term forecasts of a particular forecast aspect are generated using forecast sensitivities of that forecast aspect with respect to model initial conditions. The ensemble forecasts are generated using the differences in MM5 initial conditions derived from different analyses. The product of the differences in analyses with the sensitivities determines the approximate changes in the response function at the end of the forecast interval. The changes in the response function obtained from this procedure allow bounds on the forecast aspect to be established. Integrations of the non-linear model forward in time are used to evaluate the actual change in the response function due to the different initial conditions and to determine the validity of the tangent-linear assumption.
A statistical analysis of the results of several weeks of running this ensemble forecast generation procedure are presented and suggestions on how this procedure might be employed in real-time forecast preparation are discussed.
Supplementary URL: http://helios.aos.wisc.edu
Joint Poster Session 1, Ensemble Forecasting and Other Topics in Probability and Statistics (Joint with the 16th Conference on Probability and Statistics in the Atmospheric Sciences and the Symposium onObservations, Data Assimilation,and Probabilistic Prediction)
Wednesday, 16 January 2002, 1:30 PM-3:00 PM
Previous paper Next paper