Predictability of isotropic turbulence by massive ensemble forecasting

Turbulent flows are difficult to predict due to chaos, which amplifies any uncertainty in the initial conditions. The way this uncertainty grows and propagates is key to understanding the emergence of complexity in turbulence and to assessing the reliability of turbulence forecasts. In this Letter, a novel approach based on massive ensembles of simulations reveals that uncertainty propagates in isotropic turbulence following a simple law that depends only on the average properties of the flow. This result opens avenues to improve current forecasting techniques by efficiently and accurately modeling uncertainty propagation.