Improvement of WSR-88D VAD Winds: Cyclonic Wind Fields

Hurricanes pose a serious threat to life and property along the Gulf and Atlantic coastal regions of the United States. The WSR-88D network provides the potential to improve hurricane forecasts and warnings by monitoring changes in a hurricane's track, eye diameter, radar eyewall and rainband reflectivities. The WSR-88D Velocity-Azimuth Display (VAD) Wind Profile (VWP) display is a useful tool for diagnosis of wind fields at different altitudes as a hurricane is approaching a coastal WSR-88D.

The causes of missing winds on the VWP display were related to cyclonic flow from the approaching hurricane. The missing data arose because the extreme positive and negative Doppler velocity values around the VAD circle were inherently not 180 degrees apart and, therefore, the first-harmonic Fourier sine curve used in the operational WSR-88D VAD algorithm was a poor fit to the data. This resulted in root-mean-square (RMS) differences that exceeded the threshold value. In this situation, most of the winds on the VWP display were set equal to missing in spite of the fact that there were strong radar returns.

A new solution to recover or improve VAD winds has been developed. A higher-order polynomial regression technique employs least-squares fit of the Doppler velocity data distributed on the VAD circle. Wind speed is computed from the average of the magnitudes of the positive and negative peaks of the quasi-sine curve. Wind direction is determined from the average of the magnitudes of maximum and minimum azimuths (at which positive and negative Doppler velocity peaks occur, respectively) minus ninety degrees. After applying the experimental technique to a couple of hurricane cases such as Hurricane Katrina (29 August 2005) and Hurricane Rita (20 September 2005), the technique examines a standard deviation about a regression line which agrees well with the RMS value. The higher-order polynomial regression VAD curve fits the measurements with low RMS difference values. It is indicated that the technique does a good job of fitting the curve to the data points with low RMS difference between the curve and data points.