ICON-CH1-EPS and ICON-CH2-EPS: Probabilistic forecasts for the Alpine region
The ICON-CH1-EPS and ICON-CH2-EPS forecasting systems based on the ICON model calculate the future evolution of the atmospheric conditions for the entire Alpine region, with Switzerland in the centre of the model domain.
In contrast to the low-resolution global IFS ENS model of the ECMWF, which has a resolution of around 9 km, ICON-CH1-EPS and ICON-CH2-EPS have high spatial resolutions of 1 km and 2.1 km respectively. This high spatial resolution allows for better probabilistic forecasts for Switzerland with its complex topography, especially for extreme events such as storms or heavy precipitation.
The boundary conditions, or, in other words, the weather-related information from outside the model domain, are received by the two ensemble systems from the IFS ENS global ensemble system. The initial conditions for the forecast – the analyses – are ascertained using measurement data derived with the help of an ensemble data assimilation system. This has advantages for the model forecasts of various observation systems such as rain radar, ground-level weather stations and weather balloons.
As is clear from the acronym EPS (Ensemble Prediction System) that is contained in the name of the models, the models are calculated as ensembles. This means that several different versions of a weather forecast are generated, each of which describes a set of possible future atmospheric conditions. This helps the scientists to estimate the most probable weather development and the probability of certain weather events as accurately as possible (see also the example of probabilistic forecast products).
Why do we calculate ensembles?
Minor inaccuracies in the recording of the current atmospheric conditions can have a major impact on the reliability of weather forecasts. Even the boundary data from the global model, the ECMWF’s IFS ENS, which is used to drive ICON-CH1-EPS and ICON-CH2-EPS, have inherent errors which lead to uncertainties in the regional forecasting. Finally, the model is itself not perfect, which also leads to uncertainty in the forecasts.
The aim of ensemble or probabilistic forecasting is to represent and take these uncertainties into consideration to the greatest extent possible. To do this, numerous forecasts are calculated, each with slightly different initial and boundary conditions as well as different random variations in the model. Based on this multitude of forecasts, otherwise known as ensemble members, the likelihood of a certain event occurring can be calculated. The spread of the ensemble also provides an assessment of the predictability of the weather conditions, allowing the reliability of the forecast to be quantified.
Specifications
The high-resolution ensemble forecasting system, ICON-CH1-EPS, with its 11 ensemble members, is calculated eight times a day with a horizontal grid size of 1 km. Every three hours (00:00, 03:00, ... 21:00 UTC), the evolution of the weather is newly calculated for a forecast period of up to 33 hours. The model run that starts at 03:00 UTC has an extended forecast time of up to 45 hours, so that the forecast covers the whole of the following day. The model domain of ICON-CH1-EPS covers the entire Alpine region, with Switzerland at the centre. Vertically, ICON-CH1-EPS extends through 80 layers, up to an altitude of 22 km. The model topography reaches 4,440 metres above sea level at its highest point. At each of the 22'710'080 grid points, the weather is predicted at 10-second intervals.
ICON-CH2-EPS, the ensemble forecasting system with the longer forecast period of 8 days, consists of 21 ensemble members and is calculated four times a day (00:00, 06:00, 12:00 and 18:00 UTC) with a horizontal grid size of 2.1 km. This model domain also covers the entire Alpine region. On the vertical plane, ICON-CH2-EPS rises through 80 levels, up to an altitude of 22 km. The forecast for the 22'710'080 grid points is calculated at 20-second intervals.