Zoltan Toth
NCEP
It has been shown that information on uncertainty in weather forecasts can
be of critical value in real world applications. Therefore the transition
to probabilistic forecasting has become a priority in the US National
Weather Service. In general, probabilistic forecasts can be based either
on a single Numerical Weather Prediction (NWP) model forecast, enhanced
with corresponding verification statistics, or on an ensemble of such
forecasts (with or without statistical postprocessing). This presentation
will assess the value of the latter, ensemble approach in assessing flow
dependent variations in forecast uncertainty that cannot be easily
captured by alternative methods.