HURRICANE DIAGNOSTICS T.N. Krishnamurti Mrinal Biswas, Anu Simon,

HURRICANE DIAGNOSTICS T.N. Krishnamurti Mrinal Biswas, Anu Simon, A. Thomas ERROS IN THE CONVECTIVE HEATING IN HWRF 9mn forecast Hurricane Celia 2010 18z 22 June Inner Domain res ~ 9Km WIND PRESSURE RELATIONSHIP KATRINA 2005 Model Run at 1.33km 45 40 S u p e r g r a d i e n t W i n d s 35 30

25 20 15 10 5 0 0 0.2 0.4 0.6 0.8 Change in Liquid water mixing ratio (g/g) 1 1.2 Storm path Hurricane Celia (2010) 19-29 JUN MAX WIND (MPH) 160 Post processing of Hurricane Ensemble Forecasts T. N. Krishnamurti and Mrinal K Biswas 28 September 2010 Model Descriptions for Mesoscale Models for ensemble forecasts Models Nesting Horizontal resolution (km)

Vertical levels Cumulus Microphysic Parameterizati s on PBL Land Surface Radiation Initial and boundary conditions Initialization HWRF 2 27/9 43 Simplified Arakawa Schubert Ferrier GFS NonLocal PBL GFDL Slab Model

GFS Simplified Arakawa Schubert Ferrier GFS NonLocal PBL GFDL Slab Model Schwarzkopf and Fels (1991) (longwave) / Lacis and Hansen (1974) (shortwave) Advanced vortex initialization that uses GSI 3D-var assimilation of Doppler radar data to run in development parallel. HWRF 2 13.5/4.5 42

HWRF-X HRD version of HWRF 2 9/3 42 Simplified Arakawa Schubert Ferrier GFS scheme NCEP LSM RRTM (longwave) / Dudhia (shortwave) GFS HWRF WRF ARW (NCAR) 2 12/4 36 New Kain Fritsch (12 km

only) WSM5 YSU 5-layer thermal diffusion soil model RRTM (longwave) / Dudhia (shortwave) GFS EnKF method in a 6-hour cycling mode COAMPS-TC 3 45/15/5 (15/5 km following the storm) 40 Kain Fritsch Force and restore slab land surface model

Harshvardardet et al. (1987) NOGAPS 3D-Var data assimilation with synthetic observations GFDL 3 30/15/7.5 42 Arakawa Schubert Ferrier GFS NonLocal PBL Slab Model Schwarz-kopfFels scheme GFS GFDL synthetic bogus vortex WRF ARW 2 12/4

27 Simplified Arakawa Schubert WSM5 YSU 5-layer thermal diffusion soil model RRTM (longwave) / Dudhia (shortwave) GFS (initial and boundary condition) GFS Explicit Navy 1.5 microphysics order closure (5 class bulk scheme) GFS Correlation based model ensembles Observed increment values (Lat, Lon, Int) for each lead time

Model increment forecasts (Lat, Lon, Int) for each lead time Correlation coefficients for each model for Lat, Lon, Int at each lead time Normalize the coefficients using available member models for Lat, Lon, Int at each lead time Utilize the above coefficients during the forecast phase and construct a new forecast Training phase 2008 and 2009 storm cases (Total 164 cases) The storm to be forecasted is taken out (if it is in the training period) to calculate the correlation coefficients Hurricane Ike (2008) Ike (2008) Track Errors Sep 1-14 1200 ARFS 1000

GFDL HWRF COTC 600 AHW1 400 GFD5 ENSM 200 0 CBC 12 24 36 48 60 72 84 96 108 120 Forecast Hour

Ike (2008) Intensity Errors 45 ARFS 40 GFDL 35 HWRF 30 Error (kt) Error (nm) 800 COTC 25 AHW1 20 GFD5 15 ENSM 10 CBC 5

0 12 24 36 48 60 72 Forecast Hour 84 96 108 120 Hurricane Earl (2010) (22 Aug-5 Sep) Earl (2010) Intensity Errors ARFS 40 GFDL 35 HWRF 30 COTC

Error (kt) 25 H3HW 20 AHW1 15 GFD5 10 ENSM 5 0 CBC 12 24 36 48 60 72 Forecast Hour 84 96 108

120 Hurricane Igor (2010) Igor (2010) Track Errors (Sep 7-21) ARFS 350 GFDL 300 HWRF COTC 200 H3HW 150 AHW1 100 GFD5 50 ENSM 0 12 24

36 48 60 72 84 96 108 CBC 120 Forecast Hour Igor (2010) Intensity Errors ARFS 35 GFDL 30 HWRF 25 Error (kt) Error (nm) 250 COTC

20 H3HW 15 AHW1 10 GFD5 5 ENSM 0 12 24 36 48 60 72 Forecast Hour 84 96 108 CBC

120 Hurricane Julia (2010) Julia (2010) Track Errors (Sep 9-24) 300 GFDL 250 HWRF COTC H3HW 150 GFD5 100 ENSM CBC 50 0 12 24 36 48 60

72 84 96 108 120 Forecast Hour Julia (2010) Intensity Errors 35 GFDL 30 HWRF COTC 25 Error (kt) Error (nm) 200 H3HW 20 GFD5 15 ENSM

10 CBC 5 0 12 24 36 48 60 72 Forecast Hour 84 96 108 120