Na#onal Hurricane Center Official Intensity Errors
Assimilate Airborne Doppler Winds with WRF-EnKF Available for 20+ years but never used in operational models due to the lack of resolution and/or the lack of efficient data assimilation methods Superobservations: 1. Separate forward and backward scans; 2. treat every 3 adjacent full scans as one fixed-space radar (translation<5km); 3. thinning ---one bin for 2 km in radial distance and 3 degree in scanning angle; 4. use medium as SO after additional QC checking These SOs are generated on flight of NOAA P3 s; transmitted to ground in real-time WRF-EnKF: 3 domains (40.5, 13.5&4.5km), 60-member ensemble (Weng and Zhang 2012 MWR)!
WRF- EnKF Performance with Airborne Vr for 2008-2010!!)# %,!!!*$ %,!!!+# %,!!!"$ %,!!!!# %!!! 1@ 1N M" MK M+ M* -JK##L!K#"#!$'!1@7!M4=/= @4>>A!!!!! @%..A!!!!! 12%!!!!!!! B4A!!!!!!! I4.%<4!!!! CD=E4F!!!! 789:45;!!! -./0!!!!!! 645.!!!!!! 2008 (26) 2009 (10) 2010 (23) Storms P3 Missions Dolly 6 Fay 6 Gustav 6 Ike 5 Paloma 3 Ana 1 Bill 4 Danny 5 Alex 1 Two 3 Earl 11 Karl 4 Richard 1 Tomas 3 GH/!!!!!!! ->4!!!!!!!?8..!!!!!! 345.!!!!!! 1%<4=!!!!!!"#$ % &!!!'# % &!!!($ % &!!!)# % &!!!*$ % &!!!+# % &!!!"$ % &!! (Zhang et al. 2011 GRL)
WRF-EnKF Performance Assimilating Airborne Vr Mean absolute track (km) & intensity (kts) error for all 2008-2010 P3 missions Interpolated WSP(t) = WSP(t) - # % $ 36h - t 36h Bias(6h) & ( ' (Zhang et al. 2011 GRL)
PSU WRF- EnKF HFIP Stream1.5 Model and Filter Configura#ons for 2012 D01: 379x244x27km D02: 304x304x9km, centered by tcvitals D03: 304x304x3km, centered by tcvitals Ø WRF V3.3.1 Ø YSU PBL Ø Monin- Obukov Surface Layer Ø PSU Cd (Green & Zhang 2012) Ø Garret Ck Ø EnKF: 60 members Ø Radius of Influence: SCL Ø Covariance relaxa#on: 0.6 43 full levels with model top at 50 mb Ø ICs & BCs: GFS analysis and its forecasts Key differences from 2011: high resolu#on (4.5 3km), more ver#cal levels (30 43), surface flux (PSU Cd, Garret Ck)
Improvement of Wind-Pressure Relationship 2008-11 from 2011 to 2012 PSU WRF-EnKF Stream1.5 system 0 20 40 60 80 100 120 140 160 920 940 960 980 1000 1020 BEST A4PS 2011system APSU Garratt(2012)
WRF-EnKF Performance Assimilating Airborne Vr Mean track (km) & intensity (kts) error for all 93 P3 missions over 2008-2012 No Bias Error of maxwsp (kts) for 2008 2012 20 18 16 14 12 10 8 6 4 2 0 79 79 76 74 66 53 37 23 Interpolated WSP(t) = WSP(t) - # % $ 36h - t 36h Bias(6h) & ( OFCL ' HWRF GFDL APSU 0 12 24 36 48 72 96 120
PSU WRF-EnKF Performance for Superstorm Sandy 60-member 3-km cloud-resolving ensemble analysis forecast from 00Z Oct 26 Best Track NHC APSU GFDL model 90 75 60 Best Track H3 NHC APSU GFDL model H2 H1 45 30 TS 15 0 00Z26 024h 048h 072h 096h 120h
PSU WRF-EnKF Performance for Superstorm Sandy EnKF analysis vs. independent observations from SFMR and flight-level obs SFMR wind speed (m/s) Flight-level q (k/kg)
PSU WRF-EnKF Performance for Superstorm Sandy EnKF analysis vs. independent observations from SFMR and flight-level obs SFMR wind speed (m/s) Flight-level q (k/kg)
PSU WRF-EnKF Performance for Superstorm Sandy 10-m maximum wind swath from the 3-km deterministic forecast Hourly 10m wind composite
PSU WRF-EnKF Performance for Superstorm Sandy 60-member 3-km cloud-resolving ensemble analysis forecast from 00Z Oct 26 TS (>32kts) and Hurricane (>64kts) Strength Wind Probability
PSU WRF-EnKF Performance for Superstorm Sandy 60-member 3-km cloud-resolving ensemble analysis forecast from 00Z Oct 26 Forecasted Probability of 96h accumulated rainfall >25&100mm
PSU WRF-EnKF Performance for Superstorm Sandy 108-h 3-km cloud-resolving deterministic precipitation forecast from 00Z Oct 26 HPC Observa#on 12Z/26-30 Oct PSU WRF- EnKF determinis#c forecast
PSU WRF-EnKF Performance for Irene (2011) 3-km cloud-resolving deterministic hindcast of precipitation from 00Z Aug 24 120h accumulated rainfall observa#on PSU WRF- EnKF 120h determinis#c forecast
Concluding Remarks Hurricane intensity forecast can be greatly improved through using advanced DA techniques and a cloud resolving NWP model with assimilation of high-resolution inner-core observations Average over nearly 100 NOAA P3 Doppler missions, the PSU WRF- EnKF forecasts with assimilation of Vr has the day 1 to 5 mean intensity forecast error 20-40% smaller than NHC official forecasts The PSU WRF-EnKF experimental system performed well for all landfalling hurricanes during 2008-2012. It also shows great promise in predicting hurricane-induced rainfalls, as well as uncertaintues Future of hurricane prediction: better inner-core observations, better data assimilation, better forecast model, better computing resources
Our Ongoing EnKF Efforts not to be Reported Today Assimilation of satellite observations for hurricane prediction (satellite derived innercore wind, cloudy radiance, GPS RO, etc.) (Jerry Zhang) Assimilation of Global Hawk observations: HAMSR, HIRAS, etc. (Scott Sieron) Fully cycled realtime WRF-EnKF hurricane prediction system for NOAA/HFIP in 2013 (Yonghui Weng) Multiscale regional ensemble reanalysis of radar and other observations of MJOS during DYNAMO (Michael Ying) Multiscale regional reanalysis for the Arctic and Tibetan Plateau regions (Hans Chen) Development of COAMPS-based EnKF system towards eventually for the fully coupled system that includes the ocean (Chris Melhauser) Development of CMA GRAPES-based EnKF system (Jidong Gao) Development of an WRF-EnKF based storm scale analysis and forecast system for China FAA (CAAC) for aviation safety (Aimin Liang) EnKF storm-scale verification (Dan Stern) and ensemble sensitivity (Erin Munsell)
Best OFCL APSU d2012 APSU FZflux 105 90 75 60 45 30 15 0 Best OFCL APSU d2012 APSU FZflux 26 27 28 29 30 31 01
Courtesy of Chris Landsea and James Franklin
Sandy 20121026H2 Flt wind speed SFMR wind speed Flt t Flt q
PSU WRF-EnKF Performance for Superstorm Sandy 60-member 3-km cloud-resolving ensemble analysis forecast from 00Z Oct 26 Flt wind speed SFMR wind speed Flt t Flt q
WRF- EnKF Performance with Airborne Vr for 2008-2012 (Zhang et al. 2011 GRL)
PSU WRF-EnKF Performance for Superstorm Sandy 60-member 3-km cloud-resolving ensemble analysis forecast from 00Z Oct 26 HWind hourly 2012 10 26 01:30 >2012 10 30 07:30 40 APSU 00Z/26 2012 10 30 07:30 >2012 10 30 07:30 10mWSP 40 40 o N 35 40 o N 35 30 30 35 o N 25 35 o N 25 20 20 30 o N 15 30 o N 15 25 o N 10 25 o N 10 5 5 85 o W 80 o W 75 o W 70 o W 65 o W 60 o W 85 o W 80 o W 75 o W 70 o W 65 o W 60 o W Hourly 10m wind composite