Global water level forecast system: integrating Earth Observation data for hydrological hazard monitoring

Authors:
Open Access
Publisher website
Google Scholar
Abstract
In response to the growing need for effective hydrological hazard forecasting under increasing climate variability, this chapter presents the Global Water Level Forecast System (GWLFS), an open-source web platform that integrates Earth Observation data with ensemble hydrological forecasts to support real-time flood risk monitoring worldwide. The system combines satellite-derived water level observations from the Hydroweb project (over 20 000 virtual stations) with GEOGLOWS ECMWF streamflow simulations, applying a discharge-to-water level transformation (DWLT) method to generate a retrospective water level simulation and a 15 day 52-ensemble water level forecast. Using monthly duration curves, DWLT statistically links simulated discharge to satellite altimetry, enabling the estimation of synthetic water level time series and high-flow thresholds based on Gumbel-distributed return periods. Transformed outputs are visualized through an interactive web application built on the Tethys Platform, offering dynamic colour-coded warnings, forecast animations, historical comparisons and performance metrics. The system architecture utilizes high-performance tools such as Polars and a microservices backend to enable scalability and rapid processing across more than 25 000 monitoring stations. Validation using the Kling–Gupta efficiency (KGE) metric demonstrates the model's reliability across diverse hydroclimatic regions.

This publication has 46 references indexed in Scilit: