1. RECENT ADVANCES IN TROPICAL CYCLONE TRACK FORECASTING TECHNIQUES THAT IMPACT DISASTER PREVENTION AND MITIGATION Russell L. Elsberry Graduate School of Engineering and Applied Sciences Department of Meteorology Naval Postgraduate School Monterey, California USA Source for most materials and illustrations on the Sixth International Workshop on Tropical Cyclones (IWTC-VI) website: hetp://severe.worldweather.wmo.int/iwtc International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

2. ORGANIZATION 1.0Importance of the track prediction 1.1Wind damage associated with tropical cyclones (Topic 0.2)* 1.2Hydrological effects (Topic 0.5) 1.3Rainfall effects (Topic 0.3) 1.4Storm tide/surge (Topic 0.4) 1.5Summary 2.0Requirements for tropical cyclone track forecast 3.0Some basic concepts in tropical cyclone motion forecasting 4.0Advances in operational track prediction 5.0Future track prediction emphases 6.0Concluding remarks *Topic numbers are from the keynote session at IWTC-VI International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

3. NEED TO IMPROVE TRACK FORECASTS FIRST PRIORITY OF PARTICIPANTS AT INTERNATIONAL WORKSHOP ON TROPICAL CYCLONE LANDFALL RELATIVE CONTRIBUTIONS OF TRACK ERRORS, INTENSITY ERRORS, AND STRUCTURE ERRORS TO DeMARIA and KNAFF TRACK PROBABILITY DISTRIBUTIONS TRACK PERTURBATIONS ONLY IWTC-VI21-30 November 2006

4. TROPICAL CYCLONE MOTION SIZE PERTURBATIONS ONLY INTENSITY PERTURBATIONS ONLY IWTC-VI21-30 November 2006

5. Effect on Storm Motion Friction due to proximity to land also induces large-scale asymmetries in surface convergence Causes a landward drift of ~ 1 m/s when storm is 150 km offshore Factor to be considered in estimating the rate of increase in wind magnitude for an approaching TC From Wong & Chan (2006) Dots denote 12-hrly TC positions

6. Terrain-induced accelerations – in the form of shear lines, reverse flow, vortices, streaks and downslope winds Fine-scale Surface Wind Features Doppler radar obs. (radial wind) during T. Maggie (1999) High speed streaks (MI, MII, MIII) and traveling vortices (A) identified above the Hong Kong International Airport -> low-level wind shear and turbulence From Shun et al. (2003) Airport

7. Hydrologic Effects from Tropical Cyclones Tides/Surge Inland Flooding Landslide/Debris Flow

8. Summary  The hydrological/hydraulic forecast depends on: – Storm Track Effects on basin location – Rainfall Rate Rate Time distribution Time distribution – Storm Surge

9. Forecasting Strategy Precipitation Hydraulic Model Hydrologic Model Tidal Model Storm Track

10. Four Top Torrential Rains Caused by Typhoon 1672 mm/24h XinLiao/Taiwan Province Typhoon Carla (6718) 17 Oct. 1967 1248 mm/24h BaiXin/Taiwan Province Typhoon Gloria (6312) 11 Sep. 1963 1062 mm/24h LinZhuang/Henan Province Typhoon Nina (7503) 6 Aug. 1975 1887 mm/24h Ali mountain/Taiwan province Typhoon Herb (9608 31-1 Sep 1996)

11. Chang et al. (1993) MWR: rain rates v.s. storm center, based on 1971-1990, 82 TCs, every 3h, 1446 cases, 22 surface stations rainfall data. 2 0 x2 0 rain maps. Climatological-statistical rainfall prediction model for TC near Taiwan

12. 9:02 UTC 10 Aug 2006 Typhoon Saomai’s reflectivity in 2.4 km (Shadow) 9-10 UTC 10 Aug 2006 one hour QPE for Saomai (Shadow) 9:02 UTC 10 Aug 2006 Typhoon Saomai’s reflectivity in 2.4 km (Shadow) 9-10UTC 10 Aug 2006 Observation of one hour precipitation (Shadow and digits-mm)

13. Quantitative Precipitation Forecasting (QPF) of Tropical Cyclone Rainfall 1. Meso Scale Numerical Model  Advanced data assimilation  Fine grid space  High Vertical and horizontal resolution  Topographic Physics  Non-hydrostatic Process 2. Statistical-dynamical Scheme  Model out put Statistics 3. Statistical Methods  Climatic background

14. Very Short Range Forecast observation (Radar + raingauge) Meso-scale NWP combination Hourly, 5km grid, 1-hr precipitation forecast up to 6 hours Radar 2.5 km grid spacing spatially continuous AMeDAS ( nearly 17km spacing ) + Raingauges of other agencies ( 5200 data ) Very Short-Range Forecast From Mannoji (2005)

15. Ocean Waves MWL Wave Setup SWL ExtremeWinds Currents Wave Runup HAT MSL datum Expected High Tide StormTide Surge after Harper (2001)

16. Highlight the progress and improvements of the storm tide models (including inundation) Objectives Developments are proposed in effective operational numerical storm tide prediction facility for forecasting the total water level by including storm surge, breaking wave set-up and astronomical tides

17. Maximum surge envelopes simulated with different typhoon tracks. (unit: cm) (a) Typhoon track used in the simulations. (b) The case in which a typhoon takes the westernmost path. (c) The same as (b) but for the easternmost path

18. 1.5 SUMMARY Warnings of a tropical cyclone disaster depend critically on the accuracy of the track forecast ● Localized wind damages, including orographic effects ● Hydrological effects (flooding, landslides, debris flow) ● Heavy precipitation location and timing ● Storm tide/surge location and timing If the track is not predicted accurately, even a perfect wind structure, precipitation field, or storm surge forcing will not result in an accurate warning International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

19. 2.0 REQUIREMENTS FOR TRACK FORECASTS Accuracy of TC path requirements depend on horizontal scale of disaster event ● Width of localized wind damage ● Scale of heavy precipitation relative to rainshed size ● Width of maximum storm surge Timing (along-track) accuracy depends on the time required for preparedness activities ● Reservoir control, NASA Space Shuttle – 5 days ● Sortie ship from port – 3-4 days ● Remove oil drilling rig personnel – 2 days ● Coastal area preparations/evacuations – 1 day International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

20. 3.0 BASIC CONCEPTS IN TC MOTION FORECASTING -- I ● First-order steering concept of small vortex advected by large- scale environmental flow ● Requirement is to observe, analyze, and predict this environmental flow ● Strength and positions of monsoon trough, subtropical anticyclone, and adjacent synoptic-scale circulations ● Recently improved TC track predictions from global/regional models ● Better use of satellite observations ● Advanced data assimilation techniques ● Improved horizontal/vertical resolution NWP models ● Improved physics representations International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

21. 3.0 BASIC CONCEPTS IN TC MOTION FORECASTING – II ● Second-order motion effect is TC vortex interaction with environment ● Requirement is an accurate specification of outer wind structure ● Beta-effect propagation due to two counter-rotating gyres ● Westward and poleward motion component ● Synthetic TC observations (bogus vortex) required ● Intensity effect on motion ● Stronger (weaker) TCs are advected over deeper (shallower) steering layer International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

22. 3.0 BASIC CONCEPTS IN TC MOTION FORECASTING – III ● TC track prediction is difficult because environment and vortex change in time and interact nonlinearly ● Time variations on synoptic, intraseasonal (10-25 d and Madden-Julian Oscillation), seasonal, interannual (El Nino/ La Nina), and multi-decadal periods ● Interactions include with asymmetric convection, boundary layer, vertical shear, etc. ● Because TC track prediction is a multi-scale, non-linear interaction problem, only numerical models have great skill International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

23. 72-hour forecast errors

25. 4.0 ADVANCES IN OPERATIONAL TRACK PREDICTION In addition to improvements in dynamical model track guidance, application of consensus track forecasting approach has contributed to reduced track forecast errors ● Topic 3.1, and Topic 3.2 reports on IWTC-VI website ● Special focus session 3a at IWTC-VI was organized to inform forecasters on how to do consensus track forecasting IWTC-VI Recommendation: Facilitate the dissemination of all TC-related NWP products so that consensus forecasting can be applied at more warning centers ● Japan Meteorological Agency has password-protected website with tracks and fields from several NWP centers ● Topic 3.2 has a long list of global/regional models that might be used International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

26. Western North Pacific (1992-2006) 72-h TC Track Forecast Error (nm)

27. TOPIC 3 SUMMARY REPORT OPTIMAL USE OF GUIDANCE - II POTENTIAL REDUCTION IN 120-H ERRORS FROM OPTIMUM USE OF GUIDANCE FOR 38 (33) LSLE CASES DURING 2004 (2005) SELECTIVE CONSENSUS (SCON) HAS POTENTIAL TO IMPROVE TRACK FORECASTING WHEN ONLY A SMALL NUMBER OF MODEL TRACKS ARE AVAILABLE AND ONE OR TWO TRACKS ARE CLEARLY ERRONEOUS IWTC-VI21-30 November 2006 SCON vsCONWSCON vsJTWC Year2004200520042005 Average239222203382 Maximum6266299321256 Minimum1364-8333

28. 5.0 FUTURE SHORT-TERM TRACK PREDICTION EMPHASES Forecasters at IWTC-VI stated further short-term track forecast improvements are required for better warnings of TC landfall impacts ● High temporal resolution TC center fixes ● Vortex wind structure and precipitation observations are needed, along with advanced data assimilation ● High-resolution mesoscale numerical models including air-sea-land coupling ● Nowcasting techniques that update the longer track guidance with local observations International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

29. Average track forecast errors (nm) for the 1 to 5 day period before landfall (2001-2005)

30. 5.0 FUTURE LONG-TERM TRACK PREDICTION EMPHASES Further improvements at longer forecast intervals are possible ● Advancements in NWP model guidance via observations, data assimilation, and improved numerical models ● Advanced global models are being developed in several nations that could also be used if TC initial conditions were added ● Addition of more models that have skill, and optimum use of model guidance ● Characterization of when each model is likely to be erroneous is necessary for optimum use of consensus forecasting International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

31. 5.0 FUTURE TRACK EMPHASIS - CONFIDENCE MEASURE-I Especially longer-term track guidance can vary widely at times ● Small track guidance spread generally means high confidence ● Large track guidance spread ● May indicate large uncertainty in this forecast situation ● May indicate an erroneous track due to initial conditions or poor model forecast ● May have compensating track guidance errors that cancel in the consensus approach ● Worst-case scenario is that all track guidance is wrong, which requires more skillful models International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

32. Consensus Track Forecasting Confidence Level Small Vs Large spread in models

34. 5.0 FUTURE TRACK EMPHASIS - CONFIDENCE MEASURE-II An ensemble prediction system (EPS) is another tool to estimate track uncertainty for difficult situations ● Average of EPS member tracks is not better than deterministic model ● EPS numerical model has degraded horizontal resolution ● As of 2005, track spread from ECMWF or NCEP EPSs was not a good indicator of track forecast error ● Post-processing of EPS tracks is required ● Future opportunities from several nation’s EPS and from THORPEX Interactive Grand Global Ensemble (TIGGE) program IWTC-VI Recommendations: ● Facilitate dissemination of ensemble forecasts ● Develop probabilistic forecast products International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

36. 5.0 FUTURE TRACK EMPHASIS – TARGETED OBSERVATIONS Topic 3.3 at IWTC-VI summarized use of targeted aircraft dropsonde observations to improve TC track predictions ● USA has demonstrated value of targeted observations ● Dropwindsonde Observations for Typhoon Surveillance near Taiwan Region (DOTSTAR) program has also demonstrated value Four targeting strategies have been used to determine the most sensitive areas for deploying dropwindsondes ● Working Group formed at IWTC-VI to perform a detailed evaluation of the various techniques International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

37. CONCLUSIONS -- I ● Importance of TC track forecasts for TC disaster warning process has been emphasized ● Dramatic improvements in 72-h track forecasts have been achieved from improved numerical guidance and consensus approaches ● Extensions to 120-h forecasts have been made – as accurate as 72-h forecasts of a decade ago Major IWTC-VI Recommendations: ● Maintain and increase remotely sensed observations ● Insert TC initial conditions in other global models ● Transfer this improved technology to developing nations International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007

38. CONCLUSIONS – II ● Forecast community stated requirement for improved landfall position forecasts ● Requires high temporal resolution center fixes, advanced data assimilation techniques, higher- resolution models ● Likely future focus will be on probabilistic forecasts to assist in tropical cyclone disaster risk assessment International Training Workshop on Tropical Cyclone Disaster Reduction Guangzhou, China 26 March 2007