The problem of scaling: native precision and international attain
Specialised hyperlocal early warning methods are designed to take care of rainfall-induced flash floods in particular city environments, similar to Florida (USA), Barranquilla (Colombia), Manila (Philippines), Nakhon Si Thammarat (Thailand), Mayaguez (Puerto Rico), and Barcelona (Spain). These methods usually depend on networks of bodily sensors that monitor variables similar to direct precipitation or radar-inferred precipitation, water degree, and circulate velocity. Though extremely correct in particular areas, it’s troublesome to scale attributable to excessive {hardware} implementation prices and the necessity for site-specific tuning algorithms and engineering experience.
At a broader degree, initiatives such because the WMO’s Flash Flood Steerage System (FFGS), the Climatology-Based mostly European Runoff Index (ERIC) Flash Flood Indicator, and the US Nationwide Climate Service’s (NWS) Flash Flood Warning System present broader protection by means of distant sensing and numerical climate fashions. Nonetheless, these methods face important hurdles in international adoption. The principle downside is the reliance on high-resolution hydrological maps and radar-based climate forecasts, sources which are largely unavailable within the World South. Moreover, the reliance on skilled hydrologists to interpret advanced mannequin information and disseminate actionable warnings poses a second main problem.
To attain near-global protection, our mannequin makes use of solely international climate merchandise (NASA IMERG, NOAA CPC), in addition to real-time international climate forecasts from the ECMWF Built-in Forecast System (IFS) high-resolution (HRES) atmospheric mannequin and an AI-based intermediate-range international climate forecast mannequin by Google DeepMind. Presently, the system operates at a spatial decision of 20×20 kilometers, a constraint primarily decided by the decision of knowledge sources accessible worldwide.
