EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Tropical Cyclones in European Seasonal Forecast Models

Kevin Hodges1, Daniel Befort2, and Antje Weisheimer3
Kevin Hodges et al.
  • 1University of Reading, Dept. of Meteology, UK
  • 2AOPP, Department of Physics, University of Oxford, UK
  • 3NCAS-Climate, Department of Physics, University of Oxford, UK and ECMWF, Reading, UK

This study assesses the representation of Tropical Cyclones (TC) in an ensemble of seasonal forecast models from five different centres (ECMWF, UK Met Office, DWD, CMCC, Météo-France). Northern Hemispheric Tropical Cyclones are identified using a widely applied objective Tropical Cyclone tracking algorithm based on relative vorticity fields. Analyses for three different aspects are carried out: 1) assessment of the skill of the ensemble to predict  the TC frequencies over different ocean basins, 2) analyse the dependency between the model's ability to represent TCs and large-scale biases and 3) assess the impact of stochastic physics and horizontal resolution on TC frequency.

For the July to October season all seasonal forecast models initialized in June are skilful in predicting the observed inter-annual variability of TC frequency over the North Atlantic (NA). Similarly, the models initialized in May show significant skill over the Western North Pacific (WNP) for the season from June to October. Further to these significant positive correlations over the NA, it is found that most models are also able to discriminate between inactive and active seasons over this region. However, despite these encouraging results, especially  for skill over the NA, most models suffer from large biases. These biases are not only related to biases in the large-scale circulation but also to the representation of intrinsic model uncertainties and the relatively coarse resolution of current seasonal forecasts. At ECMWF model uncertainty is accounted for by the use of stochastic physics, which has been shown to improve forecasts on seasonal time-scales in previous studies. Using a set of simulations conducted with the ECMWF SEAS5 model, the effects of stochastic physics and resolution on the representation of Tropical Cyclones on seasonal time-scales are assessed. Including stochastic physics increases the number of TCs over all ocean basins, but especially over the North Atlantic and Western North Pacific.

How to cite: Hodges, K., Befort, D., and Weisheimer, A.: Tropical Cyclones in European Seasonal Forecast Models, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-2552,, 2020

This abstract will not be presented.