Ensemble Forecasting

  • a method used to improve medium and long-range forecasts
  • produce a model forecast from the same model or a couple models many times
  • the initial model conditions are slightly different for each run to represent the "uncertainties" and "errors" inherent in the observations used to initialize the model
  • if, at the end, all model runs produce a similar forecast, then a forecaster can have greater confidence in the model prediction.

EXAMPLE DATA:

NCEP (NMC) Ensemble Forecast Information

The following text is from : http://www.cdc.noaa.gov/~map/maproom/ENS/ens_desc.html

NCEP produces a set of 17 global forecasts each day from nearly identical initial conditions. The differences between the initial conditions are assumed to be within the range of analysis error. The set of 17 forecasts is referred to as the "ensemble forecast". The mean of the ensemble should, on average, have higher skill than any individual member. Plots of the ensemble mean 500 mb height, Mean Sea-Level Pressure and 500-1000 mb thickness for each forecast time are included, along with "spaghetti" plots which consist of a single contour of 500 mb height for each member of the ensemble. Since the spread among ensemble members is often inversely proportional to skill, the "spaghetti" plots are useful in assessing the utility of the ensemble mean.

Learn more about ensemble forecasting from the Climate Diagnostics Center: http://www.cdc.noaa.gov/spotlight/12012001/