Please note that this product is experimental and is not operationally supported, meaning that there may be some gaps in the availability on the EFAS web interface due to outages.
The radar-based intense precipitation hazard forecasts layer shows a weather radar-based nowcasting product to highlight areas (shown as ellipses) where intense precipitation may cause flash flood impacts up to 3 hours ahead. Each ellipse is shaded according to the potential level (yellow, orange or red) of the flash flood hazard, with red being the highest level. Grey ellipses show where no flash flood hazard is expected, white ellipses are where the flash flood hazard level could not be computed. The ellipses begin to fade out with increasing lead time (typically after 2-3 hours lead time), representing the spread of the ellipses as the forecast lead time is increased. An example of the layer is given in Figure 1 below, the top row shows the forecast at T+0 hours lead time, the bottom row shows the same forecast at T+2 hours lead time.
The layer was originally developed by FMI (Finnish Meteorological Institute) during the EDERA project.
Forecast at T+0 hours
|
|
Forecast at T+2 hours
|
Figure 1. Example of the Radar-based precipitation hazard layer shown on the EFAS web interface for a forecast on the 11th February 2025 at 09:00 UTC.
Methodology
The radar-based intense precipitation hazard forecast products combine cell-based storm nowcasts (Rossi et al. 2015) with a machine learning (ML) model to predict hazard level following the concepts described by Rossi et al. (2013) and Tervo et al. (2019). The product generation process is illustrated in Figure 2.
Storm cells are identified and tracked from OPERA radar reflectivity composites where the reflectivity exceeds a value of 35 dBZ (step 1).
Hazard levels of the cells are then predicted with the ML (Machine Learning) model which is trained on observations of flash flood impacts, radar reflectivity and meteorological information (specifically the Convective Available Potential Energy - CAPE, and Convective INhibition - CIN). Ellipses are fitted to the storm cells, and probabilistic nowcasts of the storm ellipse locations over the next 3 hours are produced by applying the Kalman filter-based model (step 2).
The predicted hazard levels are combined with the storm locations, which gives the precipitation hazard nowcasts (step 3). Here we assume that the hazard levels remain constant during the forecast time range.
Figure 2. Workflow to generate the radar based precipitation hazard layer.
References
Rossi, P.J., Hasu, V., Halmevaara, K., Makela, A., Koistinen, J., Pohjola, H. 2013. Real-time hazard approximation of long-lasting convective storms using emergency data. Journal of Atmospheric and Oceanic Technology. 30, pp 538-555. https://doi.org/10.1175/JTECH-D-11-00106.1
Rossi, P. J., V. Chandrasekar, V. Hasu, and D. Moisseev, 2015: Kalman Filtering–Based Probabilistic Nowcasting of Object-Oriented Tracked Convective Storms. J. Atmos. Oceanic Technol., 32, 461–477, https://doi.org/10.1175/JTECH-D-14-00184.1.
Tervo, R., Karjalainen, J., and Jung, A. 2019. Short-Term Prediction of Electricity Outages Caused by Convective Storms, IEEE T. Geosci. Remote Sens., 57, 8618–8626, https://doi.org/10.1109/TGRS.2019.2921809