15A.3 Diagnostics and verification of the tropical cyclone environment in regional models

Friday, 20 April 2012: 11:00 AM
Champions DE (Sawgrass Marriott)
Brian D. McNoldy, CIRA/Colorado State Univ., Fort Collins, CO; and K. D. Musgrave and M. DeMaria

Regional dynamical models are constantly being updated with improved physics, microphysics, thermodynamics, and increased resolution. Despite this, statistical-dynamical models such as LGEM and SHIPS continue to produce lower average intensity errors at nearly all lead times. These statistical-dynamical intensity models require current and forecast positions of a tropical cyclone, as well as current and forecast large-scale storm-centered environmental fields such as vertical shear. Statistical relationships between all of those parameters can be utilized to produce a single value: the storm intensity.

In this study, the large-scale environments of three regional hurricane models (HWRF, GFDL, and COAMPS-TC) will be verified against GFS analyses for the entire 2011 Atlantic hurricane season. These comparisons are facilitated by simple diagnostic text files in which hundreds of gigabytes of model output per run are condensed down to a 20kb text file containing various storm-centered environmental parameters. Since 2008, these diagnostic text files have been produced and posted online in real-time at CSU/CIRA (for HWRF and GFDL only). They contain six-hourly five-day forecasts of intensity, pressure, shear magnitude and direction, storm speed and heading, SST, distance to land, as well as a full sounding that includes temperature, relative humidity, geopotential height, and zonal and meridional winds. With the aid of these diagnostic text files, an entire season's worth of key vortex and environment data from a model can be stored in about 8Mb instead of 2Tb. During the 2011 hurricane season, this diagnostic file code was made available to the community so modeling groups could produce their own diagnostic files in real-time.

Errors and biases in the environmental fields may shed light on intensity errors and biases. If the model's track and/or environment are wrong, there is no hope for any amount of resolution in the inner core to accurately predict intensity.

DISCLAIMER: The views, opinions, and findings in this report are those of the authors and should not be construed as an official NOAA and/or U.S. Government position, policy, or decision.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner