Extended Abstract (148K)13B.6
A Summary of Ceiling Height and Total Sky Cover Short-Term Statistical Forecasts in the Localized Aviation MOS Program (LAMP)
Mitchell Weiss, WYLE Information Systems, Inc., McLean, VA; and J. E. Ghirardelli
Skillful forecasts of ceiling height and total sky cover are important to aviation forecasts. This is especially true for low ceiling heights (a major aviation hazard), where forecasts for projection hours two through six are critical. A product consisting of short-term forecasts for each weather element would therefore be an important tool. The Meteorological Development Laboratory (MDL) of the National Weather Service (NWS) is currently redeveloping the LAMP ceiling height and total sky cover statistical forecast guidance to run hourly, with one hour forecasts produced out to the 25-h projection. The LAMP guidance serves as an update to the Global Forecast System (GFS) Model Output Statistics (MOS) guidance. The LAMP regression equations for ceiling height and total sky cover categories include the GFS MOS guidance, the most current observations, advection model output, and geoclimatic variables as potential predictors. A “best” category forecast of ceiling height and total sky cover is generated from the probability forecasts by comparing the cumulative probabilities of the forecasts to previously determined probability thresholds representing each category.
MDL has initially developed LAMP guidance for the starting time of 09 UTC. In this paper, we compare the skill of the LAMP and GFS MOS categorical forecasts, and persistence. The comparison is done on an independent sample period, where the evaluation measures, the Heidke skill score (a measure of overall forecast skill) and the threat score (measure of the ability to forecast a threat event), are used. In addition, case studies are presented in which LAMP and GFS MOS best category forecasts are compared with the observations at selected local forecast sites during specific weather events. Initial comparisons of Heidke skill scores for both ceiling height and total sky coverage show the LAMP guidance to be significantly better than the GFS MOS for projections 1-h through 9-h and a small improvement in skill for the 10-h through 25-h projections. Threat scores for low ceiling height show significant improvement over the GFS MOS for projections 1-h though 12-h and some improvement for the remaining projections.
Session 13B, Statistical Modeling
Thursday, 4 August 2005, 1:30 PM-3:00 PM, Empire Ballroom
Previous paper
Recorded presentation