HRD Seminar – Dr. Andrew Hazelton, NOAA/GFDL, Princeton, NJ – 29 May 2018

Dr. Hazelton presented a seminar titled “High-Resolution fvGFS Forecasts of Atlantic Tropical Cyclones: Structural Analysis and Evaluation of the 2017 Atlantic Hurricane Season”.


As numerical models reach finer horizontal and vertical resolution, evaluation of storm structure is critical to model assessment and improvement.  Structural metrics based on horizontal wind are examined in high-resolution TC forecasts from the Geophysical Fluid Dynamics Laboratory (GFDL) fvGFS model.  The model forecasts are compared with aircraft radar data from NOAA P-3 flights into 7 different hurricanes from 2010-2016.  This radar comparison is a unique way to use observed TC structure to verify a model beyond the traditional track and intensity metrics.  Four different structural metrics are evaluated, including radius of maximum wind (RMW), the vertical depth of the TC vortex, the vertical slope of the RMW, and the horizontal decay rate of the TC wind with increasing radius from the TC center. 

Hurricane Matthew (2016) is analyzed in detail through comparisons of high-resolution forecasts with slightly different initial conditions and physics configurations.  The impacts of the convective scheme on the track differences are especially pronounced.  Vortex tilt is found to be an important discriminator between forecasts that more correctly predicted rapid intensification of Matthew and those that did not.  A possible physical mechanism for the (unrealistic) interruption of intensification seen in the fvGFS and some operational model forecasts is also examined. 

A similar version of the fvGFS model is evaluated in a large set of forecasts from the 2017 Atlantic Hurricane Season, including high-impact hurricanes such as Harvey, Irma, Maria, and Nate.  The model’s ability to predict the structural changes leading to the rapid intensification (RI) of Harvey is evaluated.  This analysis highlights the ways in which TC structure can be used to assess and improve TC forecasts for better prediction of the many coastal and inland hazards associated with landfalling TCs.

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