Use of radar data for TC initialization and predictions

Hurricane track forecasts improve steadily over the last several decades. However, intensity forecasts show very little improvement over the last decade because small-scale inner-core circulation can not be properly resolved by satellite observations. To improve intensity forecasts, it is important to use high-resolution Doppler radar data to properly initialize inner-core circulation for high-resolution numerical models. A technique is developed for initialization of a hurricane vortex using horizontal velocities through a deep layer of the atmosphere obtained from Doppler radar. The technique uses two new innovations. The first is the use of the mesoscale vorticity equation to diagnose the vertical velocity and divergent wind. The second is the use of the Bounded Derivative Initialization to obtain two dynamic constraints, one each for gravity and sound waves. With the fast waves controlled, a nonhydrostatic model can be initialized to allow a smooth and balanced start. The technique is tested using the 1.5 km resolution MM5 model with radar data from Hurricane Danny as it approached the Gulf Coast in 1997. Numerical results show positive radar data impacts on track and intensity forecasts. The initialization scheme correctly inserts the observed storm derived from radar data in the right location. After initialization, the forecast storm track closely follows the actual storm track observed from Doppler radar. Comparisons of forecast winds and radar wind measurements show the use of radar data substantially improves the intensity and horizontal structure of forecast wind fields. We also investigated the importance of cloud microphysical parameterizations to hurricane forecasts. Numerical experiments show that in order to improve hurricane cloud and precipitation forecasts, it is important to fine tune cloud microphysical parameterizations to account for larger drop size distributions typically found in tropical weather systems such as hurricanes.