The Impact of FORMOSAT- 3/COSMIC GPS RO Data on Typhoon Prediction UCAR: Y.-H. Kuo, T. Iwabuchi NCAR: H. Liu, W. Wang, X. Fang, Z. Ma, Y.-R. Guo CWB: C.-T.

The Impact of FORMOSAT-3/COSMIC GPS RO Data

on Typhoon Prediction

UCAR: Y.-H. Kuo, T. IwabuchiNCAR: H. Liu, W. Wang, X. Fang, Z. Ma, Y.-R. Guo

CWB: C.-T. Terng, J.-S. Hong

COSMIC Accomplishments• Yes, it is nice that we are improving operational

global model predictions, in terms of 100 mb temperature errors and 500 mb height anomaly correlations.

• Yes, it is nice that COSMIC is shown to be a very valuable data set for monitoring climate change, and calibrating other satellite measurements.

• Yes, it is nice that COSMIC data is shedding new lights on ionospheric features and helping with space weather forecasting.

• BUT….

What can COSMIC do about this?

Taiwan’s societal needs: Hazard Mitigation

• Taiwan is hit by typhoons 3.7 times a year (based on 20 years of statistics).

• Almost every typhoon produces significant damage.

• For hazard mitigation, evacuation, dam operation and rescue mission planning, Taiwan needs accurate prediction of typhoon in terms of track, intensity, rainfall, and wind gust forecasts at high temporal and spatial resolution.

• The rain and wind distributions can be quite different depending on the track and intensity of the storm impinging Taiwan.

• COSMIC is meant to be a “meteorological Satellite”. So, what can COSMIC do?

Challenges for Assessing the Impact of COSMIC data for Typhoons Affecting Taiwan

• Taiwan needs accurate rainfall and wind forecasts at high resolution (1 km, 1 hr)

• COSMIC GPS RO is not mesoscale observations:– Along ray scale is 250~300 km

• Taiwan’s dimension:– 350 km (N-S), 150 km (E-W)

• Large-scale global analysis does not provide a good description of typhoon vortex (in terms of intensity, size, and structure).

• Little observations are available in the vicinity of typhoons.• Topography of Taiwan can significantly affects the track of

typhoons, reducing the predictability, and complicating the impact assessment of COSMIC data.

But, there are hopes:• COSMIC GPS RO soundings can provide valuable information

about water vapor in the vicinity of typhoons (GPS RO measurements are not affected by clouds and precipitation).

• COSMIC GPS RO can improve the analysis and prediction of western Pacific subtropical high (which is crucial for track forecast).

• With advanced data assimilation techniques, COSMIC GPS RO data can be used to improve regional scale analysis and forecasts.

• Radar data assimilation can be performed at cloud-resolving resolution, building upon improved regional analysis

• Coupled with high-resolution mesoscale model, more accurate storm track, intensity, and precipitation forecast may be possible.

Factors affecting the impact of GPS RO assimilation on typhoon prediction

• Assimilation systems:– 3D-Var, 4D-Var, EnKF

• Details of assimilation systems: assimilation windows, cycling frequency, grid resolution…

• Bogus vortex implementation• Tuning of background and observation errors• Selection of observation operators for GPS RO soundings (e.g., local

refractivity, nonlocal excess phase, 2D ray tracing, …etc)• Amount of GPS RO during the assimilation period• Assimilation of other observations• It is often difficult to isolate the impact of a specific data type (e.g.,

COSMIC) in operational setting with continuous data assimilation.

Impact of COSMIC Data on Typhoon

• Genesis Stage:– Can COSMIC help improve the analysis and prediction

of the genesis of typhoons?

• Can COSMIC help improve the track and intensity forecast of typhoons, 1~5 day in advance?

• Can COSMIC help improve short-range, high-resolution prediction of rainfall and wind gusts?

• Here we show a few examples to illustrate what COSMIC could do to help typhoon predictions.

Impact of COSMIC Data on Hurricane Genesis

Example of Hurricane Ernesto (2006)

NCEP GSI/NMM system

36km 38L Ptop=50hPa;

Two cycling experiments were performed;

Ctrl: operational data without COSMIC GPS

GPS: Ctrl + COSMIC GPS

Cycling period: 00UTC 5——18UTC 10 September 2008

Typhoon Sinlaku (2008)

With COSMIC GPS Without COSMIC GPS

Continuous Assimilation over Four Days

With COSMIC GPSGPS – No GPS

With COSMIC GPSGPS – No GPS

Relative Vorticity 850 mb

Impact of COSMIC GPS RO Data on Typhoon Track and Intensity Prediction

Examples of Kalmaegi (2008)

Typhoon Kalmaegi (2008)

July 13-20, 2008

NOGPS GPS

Ensemble Ensemble meanmean

ObservedObserved

Ensemble Forecasts of Tracks (initialized at 00UTC 17 July)

NoGPS GPS

Left turning of the Typhoon is predicted with COSMIC GPSRO data.

Ensemble Ensemble meanmean

ObservedObserved45-km WRF/DART systemForecast performed with 15-km WRF – 16 ensemble members

NOGPS GPS

Ensemble Ensemble meanmean

ObservedObserved

Ensemble Forecasts of Tracks (initialized at 00UTC 17 July)

NoGPS GPS

Track error is smaller with GPSRO data.

Ensemble Ensemble meanmean

ObservedObserved

NOGPS

Ensemble Ensemble meanmean

ObservedObserved

Ensemble Forecasts of accumulated Rainfall (00UTC 17-18 July)

NoGPS GPS OBS

Ensemble Ensemble meanmean

ObservedObserved

Precipitation is enhanced with GPSRO data.

Prediction of Typhoon with a High-Resolution Mesoscale Model

Typhoon Jangmi (2008)

Super Typhoon Jangmi (2008)

Radar and Rainfall Obervations

WRF 1.33 km Movable GridInitialized with NCEP GFS analysis

Initial Condition:1200 UTC27 September 2008

WRF 1.33 km Movable GridInitialized with NCEP GSF analysis

Accumulated Rain + SLP Surface wind speed and directions

COSMIC Profile Availability

• COSMIC data is available continuously– Not like radiosonde data available only at 00 and 12Z

• Cycling 3DVAR for long data assimilation time window is one of the optimal assimilation scheme

22.5-1.5 1.5-4.5 4.5-7.5 7.5-10.5

10.5-13.5 13.5-16.5 16.5-19.5 19.5-22.5

22.5-1.5(+1d)

00 00z 00zFree forecast

DA time windowNCEP GFS for IC and LBCAvailale conventional data

Cold 3DVAR vs Cycling 3DVAR

• Significant improvement for trackforecast in cycling 3DVAR

Cold 3DVAR Cycling 3DVAR

Track Forecast Errors

• Cycling 3DVAR forecast is much better than cold 3DVAR forecast

• Forecast with COSMIC profile is better than that without COSMIC profile

WRF Experiments: Some Conclusions

• WRF Model with 1.33 km movable grid initialized with NCEP GFS analysis (which already assimilated COSMIC data) does show skills in track and rainfall forecasts.

• Vortex is too weak: Need careful work on vortex initialization.

• Need additional experiments to isolate the value of COSMIC data.

• It would be interesting to have a close collaboration between operational and research centers to study a few cases.

FORMOSAT-3/COSMIC Follow-On Mission:Why do we need it?

Typhoon Prediction: COSMIC GPS RO data has been shown to be valuable to

improve: Analysis and prediction of hurricane genesis Short-range prediction of typhoon track and intensity Coupled with high-resolution models offer chances to improve

rainfall and wind prediction 2,000 soundings per day is good, but not great. It is far

from saturation. Operational analysis already incorporated COSMIC

data, and already “implicitly” helping with typhoon prediction (e.g., NCEP, ECMWF forecasts)

Suggestions• Continue to improve the assimilation of COSMIC data

with advanced data assimilation systems (e.g., EnKF, 4D-Var).

• Improve vortex initialization procedure.• Improve high-resolution forecast models suitable for

operational use in Taiwan.• Conduct collaborative studies (with operational

centers) to more carefully assess the value of COSMIC in typhoon prediction.

• We need to work hard to make sure COSMIC continues, and to ensure GPS RO data are available for operations and research for many more years to come.