Predictability / Ensemble-Prediction
The atmosphere is a chaotic system. Tiny differences or uncertainties in the initial conditions can, after only a short time, lead to completely different results. The chaotic bahavior of the atmosphere can be investigated with ensembles, i.e. several individual forecasts. The term predictability implies the prediction of forecast uncertainty.
Current research results
The rapid growth of errors in atmospheric flow limits the usefulness of weather forecasts to a few days. In regions with convection and precipitation uncertainties grow particularly quickly. This figure shows the initial errors (red), which grow substantially after 2 hours (green) and 4 hours (yellow). Numerical experiments with the limited area model COSMO allow the investigation of growing errors and potential impacts of these errors on the large-scale circulation.
For more information see Selz und Craig, 2014
Weather-regime dependant impact of different sources of uncertainty in the convection permitting ensemble prediction system COSMO-DE-EPS of the DWD: The influence of boundary condition perturbations (red) on the precipitation spread is nearly independant of the weather situation (measured by the convective time-scale tauc). The impact of physical perturbations (black), on the other hand, is much larger in regimes, characterized by long convective time-scales, i.e. locally triggered convection.
For more information see Keil, Heinlein und Craig, 2013