Improvements of warm-season convective wind forecasts at the Kennedy Space Center and Cape Canaveral Air Force Station

Thumbnail Image
Loconto, Andrew N.
The purpose of this thesis was to evaluate current convective wind gust forecasting tools and develop both an extensive climatology of warm-season convective wind gusts and some new convective wind gust forecasting techniques for potential operational use at NASA's Kennedy Space Center and Cape Canaveral Air Force Station (KSC/CCAFS). The primary dataset used was an 11-year (1995-2005), May through September series of 5-minute averaged peak wind speeds from an extensive instrumentation network of towers on the KSC/CCAFS complex. After rigorous manual and automated quality control routines were performed on the dataset, a chronological dataset of "convective wind periods" --periods in which a peak wind gust was recorded when a thunderstorm(s) was occurring over a length of not more than 6 hours-- was compiled. Climatological statistics were then computed for the 11-year warm-season convective wind periods. After dividing up these convective wind periods into "KSC warning-criteria" --convective wind periods where peak wind speeds were ≥ 35 knots and "below-criteria" periods, an evaluation of current forecasting aids using RAOB and Doppler radar data for a pool of convective wind periods was performed. The RAOB data used in this study came from the most recent local KXMR sounding for any convective wind period, while the Doppler radar data utilized in this study was from the NCDC's archive of the Melbourne, FL (KMLB) "Storm Structure" datafile from 5 minutes prior to the occurrence of the maximum peak wind gust for any convective wind period. This study found Probability of Detection (POD) and False Alarm Rate (FAR) values of roughly 50% for the commonly-used binary (yes/no) RAOB-based convective wind forecast aids, MDPI and WMSI, for differentiating between warning-criteria and below-criteria peak wind gusts. Root-mean-squared (RMS) errors for RAOB-based peak wind gust forecasting tools (WINDEX, T1, T2 and Snyder Method) were found to be undesirably high, with RMS errors ranging between 9 to 21 knots. Using the Doppler radar data, much more accurate RMS error values of maximum peak wind speeds were reported using a 44-case dataset by using Echo Top/VIL (ET/VIL) Wind Gust Potential Equation than was reported by a previous study by Sullivan (1999). However, large RMS errors were found for wind speeds below KSC warning criteria. It was also found that cell-based VIL and maximum reflectivity were found to have the best correlation to peak convective wind gust speeds. Finally, work was done to develop new convective wind forecasting aids using RAOB sounding data and Doppler radar data. It was found that in the mean a potential differentiating factor between KSC warning-criteria gusts and below-criteria wind gusts was the lapse rate of equivalent potential temperature (theta-e). From this result, threshold values of theta-e lapse rate were established for several central Florida flow regimes as defined in Lericos et al. (2000). A multiple linear regression equation was also developed using cell-based VIL, maximum reflectivity and height of the maximum reflectivity as predictors. Using a 22-case independent dataset, slightly better accuracy was found using this new equation than the ET/VIL with much more accurate values reported for below-criteria winds. It was shown that for cell cores whose maximum reflectivities reside above freezing level and that are accompanied by high values of cell-based VIL, a severe convective wind gust (? 50 knots) was found to have occurred. The conditions described for severe convective winds from this study (high VIL values accompanied by cell cores above freezing level) would seem to suggest that hail was present just prior to downburst occurrence. This is in general agreement with previous modeling studies of wet microbursts in stable lapse rate regimes.