CR 39:75-86 (2009)  -  DOI: https://doi.org/10.3354/cr00801

ABSTRACT: Simulations of Hurricane Katrina using the Weather Research and Forecasting (WRF) model on a 9 km grid over the Gulf of Mexico and the southeast United States are analyzed. Global Forecast System (GFS) analyses provided the initial fields and lateral boundary conditions (LBC) 4 times per day to drive a control simulation during the period 27 to 30 August 2005. The control captured many of the observed characteristics of Katrina. A new approach, mean signal nesting, was devised to make climate change projections of the storm for each decade in the 21st century. Mean climate change signals were extracted from A2 scenario projections of the future climate by an atmosphere-ocean global climate model (AOGCM). These signals were combined with the GFS data used in the control to create the initial fields and LBC for WRF climate change simulations. This innovative method allows the LBC to retain realistic sub-daily variability present in GFS data, but still include the climate change signal. Hurricane simulations representing the earlier decades tracked east of the 2005 trajectory, and west of that track for 4 of the 5 later decades. Sensitivity experiments suggest that anticipated atmospheric warming versus expected positive sea surface temperature trends have opposing influences on developing storms. Warming trends during the 21st century are associated with ever-increasing vertical thermal stability, inhibiting initial hurricane intensification and limiting their diameters. Eye wall wind speeds in excess of 60 m s^–1 at the time of landfall could be sustained for a somewhat longer duration in storms toward the end of the 21st century.

KEY WORDS: Climate change · Hurricanes · Weather Research Forecasting regional model · WRF