The impact of horizontal resolution on the structure

The impact of horizontal resolution on the structure and intensity of Hurricane Dennis in HWRFTM simulations Robert Rogers and Sundararaman Gopalakrishnan NOAA/AOML Hurricane Research Division Miami, FL 33149

Motivation • Improvements in intensity forecasts have lagged improvements in track forecasts Annually-averaged NHC 48 -h forecast errors from 1990 to 2005 45% reduction in 15 yr Track 17% reduction in 15 yr Intensity • Numerical model guidance can be key contributor to intensity forecasts, but it faces challenges • resolution • initial TC vortex • physical parameterizations

Motivation • Rapid intensification (RI) – defined as 30 -kt increase of peak winds in a 24 -h time period convective bursts • Enhanced IR valid 0315 UTC October 17 2005 • 30 -kt Tropical Depression 24 • RI began 14 h later, becoming Hurricane Wilma Courtesy NRL Monterey • Multiscale process • environment – O(1000 km) – troughs, anticyclones • vortex – O(1 -100 km) – VRWs, eyewall, rainbands • convection – O(1 km) – convective cores, VHTs • turbulence – O(10 -100 m) – PBL, detrainment • microscale – O(1 mm) – hydrometeor production, latent heat

Questions to consider • To get improved predictions of RI, how important is it to adequately represent (resolve) the features mentioned above? • Is it sufficient to parameterize them? • Are lower-resolution ensemble approaches preferable?

Approach • simulate with HWRFTM a case of observed RI • Hurricane Dennis (2005) • period of RI beginning at 18 UTC July 6 • run two cases • operational grid length of 27: 9 km • higher-resolution run at 9: 3 km • two-way interactive movable mesh • GFDL initial conditions, GFS forecast boundary conditions, GFDL/GFS physics • no ocean coupling, no GSI • no convective parameterization on 3 km grid • models initialized at 00 UTC July 6

Dennis Observed Track and Intensity Track Intensity 7/9 7/8 7/7 7/6 RI period 7/5

Results 27 -h animation of 850 h. Pa model-derived reflectivity TM TM 27 -9 km run 9 -3 km run

Results Intensity Track 9 -3 km 27 -9 km 7/7 Best track 7/7 7/6

Results 27 -9 km run 9 -3 km run TM TM d. BZ NOAA P-3 radar 0045 -0144 UTC 7 July kt

Results 27 -9 km run 9 -3 km run TM TM Reflectivity and MSLP d. BZ TM TM Winds kt kt

Results 27 -9 km run 9 -3 km run TM TM Reflectivity and MSLP d. BZ TM TM Vorticity X 10 -4 s-1

Summary • 9 -3 km run better able to produce vortex structures • distribution of convection • wind distribution and structure • vorticity • not sure about convective structure yet • Despite this better representation of vortex structure, intensity forecast not significantly improved • track problems • vortex initialization • physical parameterizations • still insufficient resolution

Future work • evaluate differences in convective structures between two runs • vertical motion, hydrometeor, reflectivity profiles and distributions • expand tests to more Dennis forecast cycles, more cases • perform direct and statistical comparisons of vortex and convective structures with observations • ensemble runs at 9 km vs. deterministic runs at 3 km • initial condition variations • physics diversity • try runs at 1 km

Time-height mean profiles of 1. 67 -km MM 5 simulation of Dennis convective stratiform level of peak upward motion vertical motion bursts vorticity bursts