Ensemble forecasting is a technique that acknowledges the atmosphere is a chaotic system where small differences in initial conditions can produce vastly different outcomes over the medium term (the famous "butterfly effect"). Instead of running a single weather model simulation, 20 to 50 runs are performed with slightly perturbed initial conditions, generating a "fan" of possible scenarios. Ensemble analysis provides information that a deterministic forecast cannot offer: the ensemble mean is typically more accurate than any individual member; the spread indicates uncertainty—if all members agree, the forecast is reliable; if they diverge widely, uncertainty is high. Probabilities can be calculated: for example, "70% chance of rainfall exceeding 20 mm" if 35 out of 50 members produce that result. The main ensemble systems are the ECMWF ENS (51 members, 18 km, up to 15 days), GEFS from NOAA (31 members, up to 16 days), and AEMET's γ-SREPS (20 members, high resolution for Spain). This technique has revolutionised prediction of extreme events such as hurricanes, DANAs, and heat waves, enabling warnings to be issued days in advance with quantified associated risk.