AI4Fluxes

The AI4Fluxes project aims to develop a next-generation algorithm that combines all available information sources and AI methods to estimate water (ET) and carbon (GPP/NPP) fluxes in agricultural and forestry areas using data from the Sentinel-2 (10m resolution) and Sentinel-3 (300m resolution) satellites.

The main technical challenges are:

• The generation of a large database and existing ET and GPP/NPP products, and the exploitation of all this data through AI models to provide accurate and efficient algorithms and products.
• From a technological point of view, these algorithms will allow us to estimate the variables of interest with greater precision, since they will be based on existing data and measurements, unlike the physical models currently used, which have greater uncertainty associated with the model and its input parameters.
• They will also reduce the production costs of these products, since AI-based algorithms will be much more computationally efficient, since current algorithms require many inputs and calculations to estimate the final product. This is a current need in the company whose main business line is based on evaluating products generated by Earth observation services (e.g., Copernicus) where productivity (GPP) is one of them, and a agriculture business line currently in pre-commercialization phase which is focus on providing irrigation advice via satellite based on the actual ETestimate, which can be accessed through our agriService geoportal.