PyRainWarn: pySTEPS-BE ensemble nowcasts for extreme rainfall warnings in Belgium

Heavy rainfall followed by flooding is one of the most damaging weather phenomena in Belgium and Western Europe. Therefore, accurate forecasting of such extreme precipitation events is crucial for effective disaster preparedness and mitigation.

The Royal Meteorological Institute of Belgium (RMI) operates pySTEPS-BE, a seamless ensemble nowcast system built with the open-source and community-driven pySTEPS library [2]. PySTEPS-BE uses radar-based QPE and ALARO-AROME NWP input to produce 6-hour ensemble forecasts via scale-dependent blending and noise modeling, following the STEPS methodology [2,3].

PyRainWarn translates these probabilistic nowcasts into warnings at high resolution, considering uncertainties, which is not possible using deterministic nowcasts. Probabilities of exceeding rainfall thresholds are defined as the proportion of ensemble members exceeding a reference threshold. The threshold values are calculated for four return periods and four accumulation durations based on spatial generalized extreme value models [4]. They are compared to each of the pySTEPS-BE members to obtain the exceedance probabilities per pixel of the rainfall nowcast, and further processed on the administrative scale of municipalities. Finally, each municipality is colored based on 4 warning levels that are defined with the severity, i.e. the return period, and the probability, i.e. the likelihood of the event.

We present the theory and methods behind these novel extreme rainfall warnings, as well as a demonstration of the interactive map output and some actual examples. The product is currently being developed for expert users such as hydrology and emergency services, but we plan to condense and simplify the output and make it available to the general public.

 

References

[1] Pulkkinen, S., Nerini, D., Pérez Hortal, A. A., Velasco-Forero, C., Seed, A., Germann, U., and Foresti, L (2019), Pysteps: an open-source Python library for probabilistic precipitation nowcasting (v1.0), Geosci. Model Dev., 12, 4185–4219, https://doi.org/10.5194/gmd-12-4185-2019.

[2] Bowler, N.E., Pierce, C.E. and Seed, A.W. (2006), STEPS: A probabilistic precipitation forecasting scheme which merges an extrapolation nowcast with downscaled NWP. Q.J.R. Meteorol. Soc., 132: 2127-2155. https://doi.org/10.1256/qj.04.100.

[3] Imhoff, Ruben O., et al. (2023): Scale‐dependent blending of ensemble rainfall nowcasts and numerical weather prediction in the open‐source pysteps library, Quarterly Journal of the Royal Meteorological Society 149.753 (2023): 1335-1364.

[4] Van de Vyver, H. (2012): Spatial regression models for extreme precipitation in Belgium, Water Resour. Res., 48, W09549, https://doi.org/10.1029/2011wr011707.