Impact of GPS RO Data on the Prediction of Tropical Cyclones

Impact of GPS RO Data on the Prediction of Tropical Cyclones Ying-Hwa Kuo, Hui Liu, UCAR Ching-Yuang Huang, Shu-Ya Chen, NCU Ling-Feng Hsiao, Ming-En Shieh, Yu-Chun Chen, TTFRI

Outline Tropical cyclone prediction: A brief review Impact of GPS RO data on analysis and prediction of tropical cyclones: Track, intensity, rainfall, and genesis Factors affecting the use of RO data on tropical cyclone prediction: Observation operator, data QC, Obs error specification Outlook for FORMOSAT-7/COSMIC-2

Hurricane Joaquin (2015) NCEP/GFS ECMWF

Hurricane Joaquin (2015) NCEP/GFS ECMWF

Five Necessary Conditions for Tropical Cyclogenesis 1. Sea surface temperature above 26.5 27 o C 2. A deep surface-based layer of conditional instability 3. Enhanced values of cyclonic lowlevel absolute vorticity 4. Organized deep convection in an area with large-scale mean ascent and high midlevel humidity; and 5. Weak to moderate vertical wind shear Tory and Frank (2010), based on a revision of Gray s (1968) first global TC genesis climatology NOGAP T72 failed minus success TC Forecast, 500-700 hpa RH for NW Pacific storms 1997-1999. Cheung and Elsberry (2002)

Skill of Tropical Cyclogenesis Prediction by NCEP and ECMWF NCEP/GFS PD: 0.15 FA: 0.65 From (Halperin et al 2013) ECMWF PD: 0.14 FA: 0.30

4-Day Ernesto Forecasts with WRF-ARW The Actual Storm Forecast with GPS Forecast without GPS Liu et al. (2012, MWR)

Low-level moisture change by assimilating GPS Assimilation of 1 GPS RO sounding in the vicinity of Hurricane Ernesto caused a 1.8g/kg increase in moisture at 1.5 km. This was enough to get the hurricane genesis going.

48h-forecast of SLP (starting at 00UTC 25 August) August 2006 OBS: Observed storm CTRL: No GPS RO assimilation RO: Assimilation all RO data RO6km: No RO data below 6km RO2km: No RO data below 2km From Liu et al (2012)

Typhoon Morakot (2009) Main concern of typhoon prediction in Taiwan is flooding.

GPS No GPS 72hr FCST 24-h Rain (August 8-9 00Z) GPS No GPS OBS

Impact of GPS RO on T-PARC Typhoon Track Prediction SINLAKU 48-h track forecast errors, HAGUPIT averaged over 52 runs for three typhoons (Sinlaku, Hagupit, Jangmi): GPS: 104.1 km NOGPS: 137.7 km 24% improvement JANGMI ALL 3 TCs Based on WRF/DART EAKF data assimilation system

Impact Evaluation of GPS RO Data on 327 Typhoon Forecasts over 11 storms 12-h partial cycling data assimilation with the CWB operational system mean Chen et al. (2014) 95% confidence interval of Student s T distribution

GPS RO assimilation strengthens the Western Pacific Subtropical High 1 2 Averaged over 327 initializations for 11 storms

Typhoon Nuri (2008) Formed at 1800 UTC 16 August 2008 over Western Pacific Ocean. WRF Forecasts from 1800 UTC 14 August 2008 with either ECMWF or NCEP global analysis fail to predict the genesis of Nuri. Perform 3-Day data assimilation with and without the use of GPS RO data, starting at 1800 UTC 11 to 1800 UTC 14 August 2008. RO Assimilation uses nonlocal excess phase (EPH) observation operator developed by Sergey Sokolovskiy et al (2005). 4.5 Day WRF forecasts begin at 1800 UTC 14 August 2008.

WRF (15 km) Forecast initialized at 1800 UTC 14 August 2008 GFS I.C. No Genesis ECMWF I.C. 30 h delay 48h FCST Valid at 1800 UTC 16 August 48h FCST Valid at 1800 UTC 16 August 102h FCST Valid at 0000 UTC 19 August 102h FCST Valid at 0000 UTC 19 August SLP + 10m wind speed

Experimental Design Model Model grids H-resolution Model top I.C. & B.C. WRF and WRFDA v3.3.1 600x400x45 15 km 30 hpa NCEP 0.5 o x0.5 o Physical parameters Microhysics Goddard microphysics scheme Longwave Radiation RRTM scheme Shortwave Radiation Goddard shortwave Surface Layer old MM5 surface layer scheme Land Surface unified Noah land-surface model Planetary Boundary layer YSU Cumulus Parameterization Kain-Fritsch scheme DA starts Forecast Genesis Forecast Time -72-66 -60-54 -48-42 -36-30 -24-18 -12-6 0 6 12 18 24 30 36 42 48 54 60 66 72.. 3 days cycling Forecast starts 2 day prior to genesis time NO GPS (GTS): assimilate operational available data at Taiwan s CWB (no radiance) GPS (EPH): assimilation operational available data + GPS RO data by using the local operator

WRF Model Forecast Starting at 18UTC 14 August 2008, Following 3-day of Data Assimilation No GPS RO Data With GPS RO Data Integrated Cloud Hyrdometeors

WRF Model Forecast After 3-day of Data Assimilation Starting at 1800 UTC 14 August 2008 No GPS RO Data With GPS RO Data 500 mb vorticity (contour) and RH (color)

WRF Model Forecast After 3-day of Data Assimilation Starting at 1800 UTC 14 August 2008 No GPS RO Data With GPS RO Data Sea Level Pressure (contour) and PW (color)

500 mb RH (color) and Vorticity (contour) No GPS RO Data With GPS RO Data T = -72 h, start of assimilation, valid at 1800 UTC 11 August 2008

500 mb RH (color) and Vorticity (contour) No GPS RO Data With GPS RO Data T = -48 h, 24 h into assimilation, valid at 1800 UTC 12 August 2008

500 mb RH (color) and Vorticity (contour) No GPS RO Data With GPS RO Data T = -24 h, 48 h into assimilation, valid at 1800 UTC 13 August 2008

500 mb RH (color) and Vorticity (contour) No GPS RO Data With GPS RO Data T = 00 h, end of 72h assimilation, valid at 1800 UTC 14 August 2008

Time-Height section of differences in water vapor (contour) and vertical motion (color) between experiments with and without GPS RO assimilation Day 1 Day 2 Day 3 Units: Vertical motion: m/s Water Vapor: Kg/kg Averaged over a 6 o x 6 o box following the 500 mb vorticity center DA begins at t = -72h DA ends at t = 0h

Time-Height section of differences in relative vorticity (contour) and vertical motion (color) between experiments with and without GPS RO assimilation Day 1 Day 2 Day 3 Units: Vertical motion: m/s Vorticity: 10-5 s -1 Averaged over a 6 o x 6 o box following the storm DA begins at t = -72h DA ends at t = 0h

How did GPS RO data impact the genesis of Typhoon Nuri (2008)? Assimilation of GPS RO results in significant moistening of tropical lower troposphere Enhanced convective instability triggers deep convection, and produces sustained strong vertical motion. Significant mid-level vorticity is created through stretching, resulting in a robust mid-level vortex with high humidity. The moist mid-level vortex plays a critical role in fostering the formation of Typhoon Nuri (2008)

5 deg. lat. Remove 16 RO soundings near the storm kill the genesis Day 1 Day 2 Day 3 Typhoon center x 5 deg. lon. Units: Vertical motion: m/s Water Vapor: Kg/kg Averaged over a 6 o x 6 o box following the 500 mb vorticity center

Statistics for 10 Typhoons over the NW Pacific, 2008-2010 Repeat the same set of experiments on ten typhoons over the North Western Pacific 2008: Kalmaegi, Fung-wong, Nuri, Sinlaku, Hagupit, Jangmi 2009: Morakot, Parma 2010: Fanapi, Megi Probability of detection is increased from 30% (without GPS RO DA) to 60% (with GPS RO DA). GPS (EPH ) NO GPS (GTS) With genesis No genesis With genesis 3 3 No genesis 0 4 Hit is counted when model produced genesis within ±24h of the observation.

Summary Prediction of tropical cyclone remains a significant challenge: Track: Improvement becoming marginal Genesis: Low probability of detection, high false alarm Intensity: Not better than statistical methods, need high res. Future: High-resolution nonhydrostatic global modeling, requiring better and more observation for initialization Research over the past 10 years have demonstrated the value of GPS RO data on tropical cyclone prediction on: Track, intensity, rainfall and genesis Moisture information derived from RO sounding in the lower tropical troposphere, in the vicinity of the storm is most valuable

Tropical lower troposphere is a challenging for RO: Large moisture variability in time and space Large uncertainties in RO measurement and retrieval Small impact on global NWP Yet, tropical lower troposphere is MOST critical for tropical cyclone prediction To optimally use RO data for tropical cyclone prediction, efforts are needed on improving: Observation operator (See S.-Y. Chen s talk F0009) Data QC, obs errors specification (Hui Liu s talk F0062 and Ming-En Shieh s talk F0094)

km 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 Heights where GPS-RO is reducing the 24hr errors 7-35 km height interval is sometimes called the GPS-RO core region. 0 1 2 3 4 5 6 7 8 9 10 FEC % Global model sees little value of GNSS RO in lower troposphere Impact in the stratosphere is small as well

Uncertainties of RO measurements in the lower tropical troposphere

RO Soundings from FORMOSAT- 7/COSMIC-2 First Launch in 24 h A factor of 10 more RO soundings over the tropics!

FORMOSAT-7/COSMIC-2 FORMOSAT-7/COSMIC-2 to be launched in Q1 of 2017 will provide GNSS RO data with: Higher quality Higher data density (a factor of 5) We can expect significant impact on tropical cyclone prediction THANK YOU to NSPO, NOAA, and Air Force!!!