Matthew Kucas Joint Typhoon Warning Center Pearl Harbor, HI
Overview Tropical cyclone track forecasting Deterministic model consensus and single-model ensembles as track forecasting aids Conveying uncertainty in forecast products Tropical cyclone intensity forecasting Statistical-dynamical ensembles and deterministic model forecasts as intensity forecasting aids Conveying uncertainty in forecast products Ensemble product recommendations Track, intensity, genesis, and structure
Tropical cyclone track forecast ensembles Ensembles significantly aid track forecasting effort Quantifying tropical cyclone track forecast uncertainty becoming increasingly important as mean forecast errors decrease Decreasing mean forecast errors DOES NOT imply higher inherent certainty in each forecast Uncertainty in individual track forecasts remains situation-based Depends on prevailing synoptic pattern and model analysis and forecast biases DoD Tropical Cyclone Condition of Readiness (TCCOR) and naval vessel movement decisions influenced by event uncertainty need better means to convey uncertainty to customers
Poorman s ensemble of tropical cyclone track forecasts generated from the following dynamical models: ECMWF UKMET JGSM NOGAPS GFS GFDN Weber (barotropic) Member spread may indicate track uncertainty, but number of members limited and confidence in component model forecasts not quantified Track forecast process: Deterministic model consensus Forecasters identify outliers and formulate subjective hedge against consensus track May 25, 2011 0000Z consensus model track forecasts and with Goerss Predicted Consensus Error (GPCE) 70% probability circles for TY 04W from the Automated Tropical Cyclone Forecasting System (ATCF) (Goerss, 2000; Goerss, 2007)
Track forecast process: Ensembles Forecasters reference ECMWF, UKMO, JENS, GFS (EnKF), and NOGAPS ensemble track forecasts. These forecasts: Provide confidence in consensus model grouping Identify possible alternate track scenarios (especially strike probabilities) Alert forecaster to potential biases in deterministic model track forecasts May 25, 2011 0000Z ECMWF 51-member ensemble tropical cyclone strike probability forecast for TY 04W from ESRL http://ruc.noaa.gov/hfip/tceps/tceps.php May 24, 2011 1800Z JENS 11-member ensemble tropical cyclone track forecasts for TY 04W from the JMA https://tynwp-web.kishou.go.jp/ (Yamaguchi et al., 2009)
Conveying uncertainty: Track forecasts Forecast track uncertainty presented qualitatively in warning text product Graphical representation, JTWC forecast error swath, based on five-year running mean forecast errors Does not convey uncertainty in a particular forecast Current ensembles insufficient for an updated approach due to late availability, differences among ensemble forecast schemes, potential data transfer problems, etc. Better approach would blend consensus / single-model ensemble spread with historical mean forecast errors and forecaster s confidence in current scenario
Intensity forecast ensembles: Forecast process applications and conveying uncertainty Most skillful automated intensity forecast guidance available to JTWC forecasters is the Statistical Typhoon Intensity Prediction Scheme (STIPS) (Knaff et al., 2005; Sampson et al., 2008) Ensemble of statistical forecasts derived from dynamical model track forecasts and predictor variables (e.g. vertical wind shear, mid-level moisture) from dynamical models One version of STIPS incorporates GFDN dynamical model intensity forecasts Forecasters consider deterministic model intensity forecast trends; single-model ensemble forecasts add little value due to low confidence in dynamical intensity forecasts May 25, 2011 0000Z interpolated dynamical model intensity forecasts for TY 04W from the Automated Tropical Cyclone Forecasting System (ATCF) Intensity forecast uncertainty qualitatively conveyed in warning text product
Recommendations Focus on tropical cyclone track prediction: Display associated intensities with each forecast track Generate strike probability clouds and identify track clusters, quantifying probabilities associated with most probable tracks, from each NUOPC ensemble model and model grouping (i.e. NOGAPS, NOGAPS+GEFS, etc.) ideally interactive Develop probability cloud to incorporate into warning discussion and graphics Derive forecast probability density for key variables from ensemble members Consider wind field ensembles for gale and storm force wind radii forecasting Produce genesis probability forecasts for designated invest areas Identify extratropical transition location and timing for ensemble members / track clusters (Hart, 2003)
References Goerss, J. S., 2000: Tropical cyclone track forecasts using an ensemble of dynamical models. Mon. Wea. Rev., 128, 1187-1193. Goerss, J. S., 2007: Prediction of consensus tropical cyclone track forecast error. Mon. Wea. Rev., 135, 1985-1993. Hart, R.E., 2003: A cyclone phase space derived from thermal wind and thermal asymmetry. Mon. Wea. Rev., 131, 585-616. Knaff, J.A., C.R. Sampson, and M. DeMaria, 2005: An operational statistical typhoon intensity prediction scheme for the western North Pacific. Wea. Forecasting, 20, 688-698. Sampson, C.R., J.L. Franklin, J.A. Knaff, and M.DeMaria, 2008: Experiments with a simple tropical cyclone intensity consensus. Wea. Forecasting, 23, 304-312. Yamaguchi, M., R. Sakai, M. Kyoda, T. Komori, and T. Kadowaki, 2009: Typhoon ensemble prediction system developed at the Japan Meteorological Agency. Mon. Wea. Rev., 137, 2592-2604.