The forecast cloud ceiling over the Shuttle Landing Facility (SLF) at Kennedy Space Center (KSC) is a critical element in determining whether a GO or NO GO should be issued for a Space Shuttle landing. However, the Spaceflight Meteorology Group (SMG) forecasters have found that ceiling is a challenging parameter to forecast at the SLF, even in the short-term (0-6 hours) when persistence is assumed to be a reliable predictor. Recent studies presented in the literature have shown success using statistical methods to improve the short-term ceiling forecast of the standard Federal Aviation Administration (FAA) Flight Rules (FR). In these studies, multiple linear or logistic regression equations were developed that used conventional surface and upper-air rawinsonde data from the forecast site as well as surrounding stations. These observations-based equations consistently out-performed the benchmark persistence climatology equations. The persistence climatology equations consist of the ceiling category observation at the initial time and a value representing the frequency of occurrence of the ceiling category at the valid time. These results were the impetus behind the Applied Meteorology Unit (AMU) task to develop short-term ceiling forecast equations for the SLF.

The AMU task differed from the studies in the literature in two ways. First, the previous studies used data from areas such as the eastern United States and the San Francisco Bay area where persistent ceilings were known to exist. Such conditions are not the norm in the subtropical environment of east-central Florida. Second, the studies used standard FAA FR cloud ceiling categories as predictands. The predictands in the AMU task were the ceiling thresholds as defined by the Space Shuttle FRs: · < 5000 ft (Return to Launch Site) · < 8000 ft (End of Mission) · < 10 000 ft (Navigation Aid Degradation)

Following the procedures outlined in the literature, the AMU used a 19-year record (1979–1997) of hourly surface cool season (October – March) observations from the SLF and surrounding stations in east-central Florida to develop observations-based and persistence climatology equations. Equations were developed for each ceiling category and each hour of the day for 1-, 2-, and 3-hour lead times. In almost every case, the observations-based equations produced an improvement over the persistence climatology forecasts. The smallest improvements were found for the 1-hour forecasts where it is assumed that persistence climatology is a strong performer. Larger improvements were produced with the 2- and 3-hour forecasts, time periods over which persistence is likely to be less of a factor. The probability of detection (POD) and false alarm rate (FAR) was calculated for each equation to determine the reliability of the forecasts. The average PODs (FARs) ranged from 84% (15%) for the 1-hour forecasts to 53% (35%) for the 3-hour forecasts.

This paper will present the data and methods used in equation development, the results of the comparisons, a discussion on the predictors chosen and their importance in the equations, and a description of possible improvements to the methods used.