PRECIPITEAU

What is Precipiteau?

Precipiteau is a real-time monitoring platform that shows how precipitation zones are affecting different regions across Europe. Its primary goal is timely awareness — helping users understand current and near-future precipitation conditions, so they can make better decisions and, where possible, avoid dangerous situations caused by heavy rain, freezing rain, or snow. The name itself unites precipitation and eau — water in French.

The platform combines live radar observations, satellite data, and numerical weather model output to provide both a current picture and a short-range outlook, updated automatically around the clock.

Data Sources

Radar map (RATE) — Near-real-time precipitation rates from the EUMETNET OPERA pan-European radar composite network, covering 35+ European countries. Composites are updated every 15 minutes and retrieved from the open openradar S3 archive.

Daily Total map (ACRR) — Accumulated precipitation since 00:00 UTC, built from the official OPERA hourly accumulation product (ACRR). Because ACRR carries an approximate 2-hour lag from real time, the most recent hours are estimated directly from the radar rate until the official accumulation files become available. The optional + Satellite mode fills radar coverage gaps using dual-satellite confirmation (H-SAF H60 + JAXA GSMaP-NOW), same blending as Radar + Sat.

Precipitation type (ICON-EU) — Surface precipitation type classification (rain, snow, freezing rain, ice pellets, mixed, hail) is derived from the ICON-EU model, operated by Deutscher Wetterdienst (DWD) and published openly every 3 hours.

Satellite precipitation (dual-source) — In Radar + Sat and Daily Total + Satellite modes, OPERA radar is blended with two satellite sources: H-SAF H60 (MSG/SEVIRI geostationary, 15-min) and JAXA GSMaP-NOW (LEO microwave, 30-min). A pixel is accepted only when both sources independently agree on rain, ensuring conservative, high-confidence estimates.

1-hour forecast (PySTEPS) — The ⚡ Forecast overlay is a nowcast generated from recent OPERA radar frames using PySTEPS optical-flow motion estimation (Lucas–Kanade). In Radar + Sat mode the input is radar–satellite blended, extending the nowcast skill into areas with sparse radar coverage. No NWP model is involved.

Methodology

Precipitation type is classified by our own implementation of the Ramer ice-fraction algorithm applied column-by-column to ICON-EU temperature, relative humidity, and wet-bulb temperature profiles downscaled to a 1 km terrain model. The method diagnoses rain, snow, freezing rain, ice pellets, and mixed precipitation at the surface.

Reference: Ramer, J. (1993): An empirical technique for diagnosing precipitation type from model output. Preprints, 5th Int. Conf. on Aviation Weather Systems, Vienna, VA, Amer. Meteor. Soc., 227–230.

Team

Precipiteau is built by a data scientist and a meteorologist who crossed paths at WeatherXM and share a deep passion for weather.

Disclaimer

Precipiteau is provided for informational purposes only. The precipitation data displayed are derived from third-party sources (EUMETNET OPERA, DWD ICON-EU, H-SAF/EUMETSAT, JAXA) and may contain gaps, artefacts, or delays inherent to radar and NWP systems. The developers make no warranties, express or implied, regarding the accuracy, completeness, or timeliness of the information presented. Precipiteau shall not be held liable for any decisions made or actions taken based on the content of this platform. Do not use this service as the sole basis for safety-critical decisions. Always consult official meteorological authorities and emergency services in situations involving risk to life or property.

News

Limitations

Radar Data

Our radar-based precipitation products use OPERA weather radar data. Although the data undergo quality control and several types of radar noise are removed, some non-meteorological echoes or residual artefacts may still remain in the final products.

Radar rainfall and precipitation accumulation are estimates, not direct measurements. They are derived from empirical relationships between radar reflectivity and rain rate. A commonly used example is the Marshall–Palmer relation, Z = 200 R^1.6, but such relationships vary with precipitation type and atmospheric conditions, so uncertainties remain.

Radar-based precipitation estimates may be less accurate both very close to the radar and far from the radar. Close to the radar, there are sampling limitations because weather radars cannot observe directly overhead at all heights. Farther from the radar, the beam rises and widens with distance because of Earth curvature, atmospheric refraction, and beam broadening. As a result, the radar increasingly samples higher parts of the cloud rather than precipitation near the ground, which can lead to larger errors in surface rain-rate and accumulation estimates.

Satellite Precipitation (H-SAF H60 + JAXA GSMaP-NOW)

In Radar + Sat and Daily Total + Satellite modes, two independent satellite sources are used: H-SAF H60 (MSG/SEVIRI geostationary, 15-min) and JAXA GSMaP-NOW (LEO passive microwave, 30-min). Both infer precipitation from measured radiation rather than observing it directly, and both carry empirical uncertainties — they may under- or overestimate precipitation rates, particularly for convective events.

A satellite pixel is accepted only when both sources independently agree on rain, ensuring conservative, high-confidence estimates. The satellite contribution is blended with radar using a distance-based weighting — radar is authoritative near active stations, with satellite filling coverage gaps at range. Both sources carry a lag of ~15–30 min, so the most recent 1–2 frames may show radar only. Uncertainties remain in regions far from any radar.

Temperature Profile & Precipitation-Type Classification

The temperature profile used in the project is based on the ICON-EU numerical weather model, which has a horizontal resolution of about 6.5 km and hourly output during the first forecast hours. To better estimate surface temperature, we adjust the model data to the terrain using a 1 km digital elevation model (DEM). However, this still provides an estimate rather than a direct ground measurement, so differences from real conditions may occur. In addition, the ICON parameters are linearly interpolated in time from hourly data to 15-minute intervals. These uncertainties may influence the classification of hydrometeor type.

The hydrometeor-type classification is based on temperature and humidity profiles from the first forecast hours of ICON-EU. These short-range forecasts are generally expected to be close to reality, but they are still model forecasts and therefore contain uncertainty, which usually increases with forecast lead time.

Despite these limitations, improving the accuracy of both temperature profiles and surface precipitation estimates remains a continuous goal of the Precipiteau team.

Contact