NOAA Privacy Policy | NWS Disclaimer
N.O.A.A. logo HWRF banner image National Weather Service logo

September 23, 2010 Meeting Summary

Jun Zhang from HRD presented his work on evaluating the planetary boundary layer (PBL) scheme in HWRF using aircraft observations. Jun first presented some of his previous work on aircraft turbulence observations. The data for this work was taken from the CBLAST experiment. Interesting results from this work included different results in Jun's vertical profile of momentum flux compared to Moss's 1978 profile. Also, the specific humidity profile was nearly constant then decreasing with increasing height while the sensible heat profile was mostly negative. Jun also noticed an imbalance in the TKE budget between the dissipation rate and shear production, which are usually balanced in non-hurricane conditions. Overall, Jun's theoretical method of dissipative heating overestimated the dissipative heating amount by a factor of three.

Jun briefly discussed his current work on evaluating mean boundary layer structure using dropsonde data. Overall, there were many dropsonde locations compared to storm center locations, meaning good coverage. Interesting results included gradient wind balance tests producing gradient wind values that were perhaps too high. Jun mentioned revisiting this work, and then presented some figures from his paper with David Nolan in Monthly Weather Review.

Jun then moved onto his current model evaluation work where he looked at diagnostics for the experimental version of HWRF using the high resolution hurricane (HRH) model runs compared to aircraft observations. 69 test cases were used from 9 storms from 2005 and 2007 (Emily, Katrina, Ophelia, Phillipe, Rita, Wilma, Ingrid, Humberto, and Karen) with 27/9 km resolution versus 9/3 km resolution. Results for tangential wind showed the 9/3 version capturing the general structure shown by observations but with too large a magnitude. The radial wind plots showed the 27/9 and 9/3 with similar structure to observations but too large in magnitude. Finally, the temperature plots showed 27/9 and 9/3 values that were very warm near the storm center compared to observations.

Jun concluded by presenting future work which will include case studies looking into PBL structure and compared to available observation data (from radar, dropsondes, and buoys). Work on composites of dynamic and thermodynamic fields from HWRF would be performed, and these composites would be compared to dropsonde and radar data. Overall, Jun wanted to improve the HWRF PBL scheme, perhaps including effects from boundary layer rolls.

Next, Zhan Zhang presented his work on the HWRF-GEFS hurricane ensemble system, specifically uncertainty in initial large scale flow. First Zhan explained his experiment structure, which included HCTL (which used HWRF V3.2 (R2) with GFS T382L64), ZPxx (which ranged from ZP01 through ZP21 for the ensemble members which used GEFS input (T190L28) with HWRF V3.2), ZEMN (the ensemble mean), and HWRF (operational HWRF using GFS T382L64, or H210 version of HWRF). Each experiment was conducted for Alex and Celia from the 2010 hurricane season. Track error plots showed a reduction in error for ZEMN compared to HCTL for both Alex and Celia, and especially later in the forecast. Intensity error plots showed almost no impact from ZEMN compared to HCTL. This was expected because only the large scale flow and boundary conditions were different for each ensemble member. No perturbations are generated inside the HWRF. Zhan then presented a figure of ensemble track spread for Alex, which increased with time, and intensity values from each ensemble member for Celia. For the intensity plot, a negative model bias was still present with the operational HWRF being the outlier with the lowest intensity early on.

Zhan then presented some rank histograms used for ensemble spread verification to see if the ensemble track improvement was due to ensemble perturbation biases. For intensity, there was a negative bias for Alex and Celia, which was produced by the model. Zhan explained that he used a bias correction on ensemble members for track latitude to reduce a northerly bias for both storms. A westerly bias was seen in the plots for track longitude, which was stronger for Celia. In the track spread versus skill relationship diagram, more spread was shown for 72h forecasts. Track probability forecast plots for Alex and Celia also showed multiple storm centers later in the forecast meaning the probability was not evenly distributed in space.

Zhan concluded by recapping his presentation and discussing his future work. This work included combining GEFS-based perturbations with physics-based perturbations and running experiments. Zhan also planned to look at flow-dependent initial position perturbations.

Please e-mail comments, questions, or suggestions about the contents of this webpage to Janna O'Connor, at

Home | HWRF Main Page
EMC | NCEP | National Weather Service | NOAA | Department of Commerce