Validation and development of existing and new RAOB based warm-season convective wind forecasting tools for Cape Canaveral Air Force Station and Kennedy Space Center
Updated forecasting methods are being constructed using open-source “R” statistical software and incorporate several classical statistical techniques such as multiple linear regression and logistic regression as well as newer techniques like classification and regression trees (CART). Validation is presently being completed on the newly developed forecasting processes with an independent dataset from the 2008 warm-season. In addition, three classes of composite 1500 UTC KXMR profiles of theta-e, temperature, dew point, relative humidity, mixing ratio, wind speed, and wind direction are being built with one class for non-convective days, a second class for days with convection and observed wind gusts less than 35 knots, and a third class for convective days and peak winds in excess of 35 knots to aid in diagnosing what meteorological parameters favor the occurrence of warning criteria winds. A previous climatology of the distribution of convective wind speeds indicated that differentiating between convective winds less than 35 knots and greater than 35 knots is important to 45 WS operations.
Although much of this research is still currently underway, preliminary results yield more promise for differentiating between convective and non-convective days than for distinguishing between convective days with wind gusts above or below 35 knots. So far, parameters showing potential for discriminating between days with warning (≥ 35 knots) or non-warning criteria winds are theta-e profile, wind speed from the surface to 500 hPa, and wind direction from the surface to 500 hPa. Days with a larger theta-e differential between the surface and the mid-levels, a southwesterly wind flow, and slightly stronger low and mid-level winds are more likely to produce warning level convective winds.