Ensemble Prediction System

It is a probabilistic forecasting method that integrates a set of simulations carried out by a meteorological or climate model, each of which starts from slightly different initial conditions. By proceeding in this way, it is possible to bound the uncertainty inherent in the future evolution of the atmospheric system, given its chaotic nature. Unlike what happens with a deterministic forecast, in which only a single solution is considered as the future state of evolution of the system, in this case we have a set of those states, which can be grouped by similarity, assigning each group a given probability of occurrence. If 50 simulations are considered and 40 of them evolve over time in a similar manner, then we can be reasonably confident (80% probability) that the actual evolution of the atmosphere will be similar to that indicated by those 40 similar future states.