The sensitivity of the Hurricane Weather Research and Forecasting (HWRF) model to four cumulus parameterization schemes was tested by the Developmental Testbed Center ( DTC) in 2012. Selected storms of the North Atlantic and Eastern North Pacific basins were run using the DTC computational environment, which is functionally-equivalent to the operational one. A control simulation using the HWRF Simplified Arakawa-Schubert (SAS) parameterization was compared against configurations using the Kain-Fritsch parameterization, the Tiedtke scheme, and a new implementation of SAS by Yonsei University.

This test used an experimental version of HWRF used in pre-implementation testing at NCEP/EMC as of February 2012. This version is intermediate between the public release versions HWRF V3.3a and V3.4a A triple-domain configuration with grid spacings of 27-, 9-, and 3-km, matching the 2012 operational HWRF implementation, was employed.

The results highlight that HWRF has substantial sensitivity to the cumulus parameterization, which can influence storm track, intensity, and size. As detailed in the Final Report, no alternate scheme provided an across-the-board substantial advantage over the HWRF SAS.