Predictability of high-impact weather events (Invited Presentation)

Predicting high-impact weather events is a crucial task for forecasting centres and is clearly a multi-scale issue. On the subseasonal time-scale, the aim is to predict flow-regimes that have higher likelihood of extreme events. In the medium-range the challenge is to predict the timing, position and duration of the event. However, for some types of extreme weather it is still difficult for current models to capture the intensity even at the shortest scales.

As extreme events by nature are rare, and each case is unique, a statistical evaluation is not straightforward. It is therefore necessary to identify key features for the development of the extreme events that are possible to verify and to identify where also less extreme cases will be useful.

For the European heat-wave in 2015, relevant factors included the signal from soil moisture in the sub-seasonal range, Rossby wave packages in the medium-range and difficulties to capture extreme temperatures at the short range. For tropical cyclone Pam, 2015, a strong MJO increased the tropical cyclone activity and this was well predicted on sub-seasonal range; in medium-range we have the ability to predict TC tracks although with some uncertainties, but on short-range we still often have the issue of predicting the intensity.

In the presentation I will give examples such as these of high-impact events to demonstrate the challenges in sub-seasonal, medium-range and short-range predictions and what type of signals are expected. For each of them I will give example of evaluation and diagnostics that can be undertaken to better understand the predictability and limitations of the capability.