HRD Seminar – Dr. Andy Hazelton, Princeton University and NOAA/GFDL – 25 October 2016

Dr. Hazelton presented a seminar on “Evaluation of Tropical Cyclone Structure in High-Resolution FV3 Simulations”.


The FV3 dynamical core with GFS physics has shown skill both in global forecasting and the ability to simulate high-impact weather events.   Here, a regionally stretched and nested version of the model, centered over the Atlantic Basin with a grid spacing of approximately 3 km, is used to evaluate tropical cyclone (TC) forecasts of track, intensity, and structure. Two different configurations are tested: one in which the convective parameterization is turned on on the 3- km nest, and one in which the convective parameterization is turned off.  In addition, a new modified version of the Lin Microphysics (called GFDL Microphysics) is tested against the two simulations using the original Zhao and Carr (1997) microphysics. Simulations are performed on 3 TCs: Earl 2010, Edouard 2014, and Gonzalo 2014

In addition to verification of track and intensity, the model is evaluated by comparison of the model data with high-resolution airborne Doppler radar composite from the NOAA-P3 aircraft.   Structural metrics analyzed include the radius of maximum wind (RMW), the slope of the RMW, and the depth of the vortex. Preliminary results indicate that keeping the convective parameterization turned on results in the best forecast of structure (particularly more realistic RMW and vortex depth) as well as a better forecast of intensity and rapid intensification.  In addition,  the  new  microphysics  scheme  appears  to  produce  more  realistic  TC  structure. Simulations of more TCs are planned in order to generalize the results and investigate further ways to improve FV3 TC forecasts.

A recording of the presentation is available on the anonymous ftp site: