Preliminary evaluation of a First Guess ceiling forecast at Space Shuttle landing sites

The National Weather Service Spaceflight Meteorology Group (NWS SMG) at NASA Johnson Space Center is responsible for issuing forecasts for all potential Space Shuttle landings. These landings include planned End-of-Mission landings at landing sites in the United States (US), emergency Trans-oceanic Abort Landings (TAL) during launch, and on-orbit contingency landings at locations throughout the world. Forecasts are issued, depending on mission requirements, for time periods ranging from 30 minutes to 5 days. One of the most critical forecast elements is the cloud ceiling. Previous studies of 12-hour to 2-day ceiling forecast skill for Space Shuttle missions shows this is an area where SMG forecasters can improve.

An automated first-guess technique for predicting ceilings using numerical model grid data has been tested at SMG. The technique uses an algorithm developed for the AWIPS Forecast Preparation System to determine cloud layers and amounts from forecast profiles of relative humidity. The US sites chosen for verification were Kennedy Space Center, Edwards Air Force Base, and White Sands Space Harbour. The TAL sites at Zaragoza and Moron in Spain, Ben Guerir in Morrocco, and Banjul in the Gambia were chosen for the overseas verification sites. Forecasts were issued twice per day for the 6- through 36-hour forecast period using the Eta model for the US sites and the Aviation (AVN) model for both US sites and the TAL sites for 45 days in January and February 2000. The forecasts were verified against the observations for the ability to predict ceilings below 8000 feet (5000 feet) for US (TAL) landing sites. Forecast ceilings at or above the threshold were considered GO forecasts. In addition, the first guess forecasts were compared to the forecasts made using the Nested Grid Model (NGM) Model Output Statistics (MOS) for the US landing sites.

The Eta first-guess correctly predicted the GO / NO-GO category 88% of the time for US landing sites with a Heidke skill score of 0.31. The AVN first-guess correctly predicted the GO / NO-GO category 87% of the time for US landing sites with a Heidke skill score of 0.28. The corresponding values for ceiling forecasts based on the MOS were 84% and a 0.32. For the TAL sites, the AVN first-guess predicted the GO / NO-GO category 94% of the time with a skill score of 0.22.

MOS is still difficult to improve upon for the US landing sites although the Eta model first-guess shows skill comparable to MOS. The AVN first-guess may be particularly useful for the TAL sites where no statistical objective guidance currently exists for Shuttle Flight rules. The use of both the first-guess ceiling forecast and MOS as forecast decision aids is recommended.