An introduction to NCEP SREF aviation project

A limitation in deterministic numerical weather forecasting is the chaotic nature of the atmosphere, which results from uncertainties in both model physics and initial conditions. One solution is the recently developed technique called the ensemble forecast. That is, a set of model runs called ensemble members are created by using (i) a single model with different initial conditions, or (ii) different models with the same initial conditions. Motivated by the success of NCEP's global ensemble forecast system, the NCEP multi-initial condition and multi-model Short Range Ensemble Forecast (SREF) system was developed in 1996 and implemented operationally in early 2001. The evaluation of SREF has shown improvements in providing CONUS forecasts during the 1-3 day time range. The SREF system is running operationally 2 times a day, producing ensemble forecasts from 15 members: 5 ETA members, 5 ETA Kain-Fristch members, and 5 Regional Spectral Model (RSM) members. To further exploit the utility of ensemble forecasting, the NCEP SREF system has been applied to aviation forecasting since late 2002. The current SREF aviation ensemble forecast has 11 primary ensemble products, including the probability, mean and spread of: icing, turbulence, cloud, ceiling, visibility, jet stream, lower level wind shear, tropopause height, etc. This paper will present an overview of the NCEP SREF aviation project, including its development plans, configurations, and algorithms. The verification problems related to ensemble forecasts will also be discussed.