Hurricane Katrina embarked upon two periods of rapid intensification (defined as a 30 kt or greater intensity increase in a 24-h period) between 26 and 28 August as it moved over the warm ocean loop current. The first period involved an increase in the maximum sustained winds from 65 kt to 95 kt in a 24-h period ending 0600 UTC 27 August. An eye became clearly evident in infrared satellite imagery early on 27 August, and Katrina became a Category 3 hurricane with 100 kt winds at 1200 UTC about 365 n mi southeast of the mouth of the Mississippi River. During the remainder of the day, the inner eyewall deteriorated while a new, outer eyewall formed, and the intensity leveled off at 100 kt. Accompanying the intensification and the subsequent deterioration of the inner eyewall was a significant expansion of the wind field on 27 August. Katrina nearly doubled in size on 27 August, and by the end of that day tropical storm-force winds extended up to about 140 n mi from the center. A cursory examination of satellite images shows a possible wind surge from the southwest which could have contributed angular momentum to the storm, in combination with the large width of the warm oceanic waters that may have enhanced convection and low-level convergence.

Although the rapid intensification of Katrina was noteworthy, the expansion of the tropical storm-force winds is the key forecast issue. The devastation wrought by this storm upon landfall is attributable more to its size rather than its intensity, as it landed as a Category 3 hurricane. This large hurricane caused a record storm surge and exposed Louisiana and Mississippi to hurricane-force winds for a long period of time.

The Hurricane Weather and Research Forecast system (HWRF) model will be used to explore the causes of this intensification and wind field expansion. HWRF is a coupled air-sea-land prediction system with a movable nested grid and physics suitable for high resolution.

Data assimilation will be conducted using a 3-h 3DVAR data assimilation cycling run. Sensitivity experiments will be performed using conventional observations, resonnaissance, satellite, and radar data. No bogussing will be used, as model initialization will be attempted using real inner-core wind data. Model coupling experiments will then be performed, with careful consideration of oceanic data in the loop current region, using the Model Coupling Environmental Library.

Once a successful simulation of the hurricane structure is completed, relative angular momentum (RAM) calculations will be performed to understand the cause of the wind field expansion. The methodology of Tuleya and Kurihara (1975) is to be followed, and the individual contributions by the following terms will be assessed: 1) mean flux convergence of RAM; 2) eddy flux convergence of RAM; 3) Coriolis contribution; 4) horizontal torque; and 5) vertical torque. The utilization of other cylindrical terms, broken up by quadrants, will also be done as necessary, such as done by McBride (1981).