Forecasting Water Quality from Space (FC-WQ)

Despite their global coverage and public availability, measurements of water quality parameters retrieved by remote sensing have not been used for local short-term forecasting of water quality yet. In fact, forecasting based on remote sensing water quality products with on-ground resolution from, say, 30 m (Sentinel-2 MSI, Landsat-8/9 OLI) to 300 m (Sentinel-3A/B OLCI) or 500 m (GOCI) constitutes a potential breakthrough since the availability of in situ measurements of water quality is often limited. For a given location, availability of remote sensing water quality products is mainly determined by satellite revisit frequencies (3-5 days for Sentinel-2A/B, 1-3 days for Sentinel-3A/B, and 30 minutes for GOCI) and constrained by cloud coverage.

The Forecasting Water Quality from Space (FC-WQ) project aims to develop and validate a method for local short-term forecasting of water quality based on EO data in coastal and inland waters. This method will be based on time series of remotely sensed water quality parameters, combined with past, present, and forecast data of physical parameters (i.e., meteorological, hydrological, and specific environmental parameters). The proposed method exploits the capability of Machine Learning (ML) to learn and model the complex relationships in aquatic ecosystems.

Specific objectives of this undertaking comprise:

• Forecasting of turbidity and chlorophyll-a concentration and associated uncertainties up to five days ahead of time
• Providing the probability of occurrence of a harmful algal bloom (HAB) within seven days ahead of time
• Development of the methodology based on selected coastal and inland water test sites while aiming at applicability to coastal and inland waters in general
• Validating the methodology for selected coastal and inland water test sites for certain hindcast periods not considered and during model training
• Demonstrating the forecasting capability at selected coastal and inland water sites for a certain forecast period
• Testing transferability of the methodology to other sites
• Developing a roadmap for further scientific studies as well as product and service applications