Initialization techniques for tropical cyclones using the Advanced Hurricane WRF

It is well documented that while great strides have been made in forecasting the tracks of tropical cyclones, relatively little progress has been made in forecasting their intensity, especially rapid intensification. Previous research has shown that the track of a tropical cyclone is largely based on its large scale flow, which is relatively well-understood. However, its intensity appears to be a function of smaller, less-understood processes. It is therefore hypothesized that a better understanding of tropical cyclone intensity change can be obtained by studying the processes occurring in the storm's core.

One of NCAR's ongoing accomplishments is the development and revisions of the Weather Research and Forecasting (WRF) computer model. This non-hydrostatic model is becoming increasingly popular for both operational and research purposes. With the multitude of options available for the Advanced Research WRF (ARW) core, the model can be tailored to suit many aspects of meteorology. Specifically, a team at NCAR has developed a set of parameters to study tropical cyclones. This Advanced Hurricane WRF (AHW) is being utilized for our ongoing research.

This project focuses on simulating real world tropical cyclones. The goal is to determine if the AHW model can adequately simulate rapid intensification and rapid weakening during a storm's lifecycle. Of specific interest is the 0 to 36 hour forecast timeframe. Simulations will be conducted at multiple stages during a storm's lifecycle utilizing TC bogussing and digital filterization techniques. The simulations will employ three domains (36 km, 12 km, and 4 km) with two-way nesting. The two inner domains will be centered on the storm's core using the vortex-following algorithm.

Since we are interested in the 36 hour forecast, the initialization of the storm is very important. There are multiple sources of input data for the WRF model, including GFS final analysis, and GFDL datasets. The proposed poster will include preliminary results detailing which dataset is best; how effective the TC bogussing and digital filterization techniques are at initializing the storm; and whether or not the best initialization yields the best forecast at 36 hours. Future research will investigate how well the model simulates intensity changes compared to the National Hurricane Center's best track.