Predictability of high-impact weather events: Sensitivity to upper-level atmospheric anomalies

As powerful as they may be, numerical weather prediction models still contain uncertainties. In particular, their ability to represent multi-scale processes remains imperfect due to the nonlinear dynamics of the atmosphere. Rather than trying to combat these uncertainties, Simon Fresnay is examining them using simulations of upper-level atmospheric anomalies, which are precursors of storms and floods. “We need to better manage uncertainties in order to better manage emergencies and know when to activate warning systems.”
One promising way to reach this aim is to simultaneously use models with slight perturbations for the same forecast. The first results on flooding in southeastern France showed that the forecast was highly sensitive to small changes to upper-level atmospheric anomalies.