In many mountainous regions like the Andes, it's hard to get accurate and useful maps of land cover due to steep terrain and limited data. This makes it difficult to monitor ecosystems, manage natural resources, or assess climate risks. Project GRADIENTES aims to change that by building a reliable, flexible model that uses satellite images from Sentinel-1 and Sentinel-2 to map land cover in the Peruvian Andes.

Leveraging an international consortium, including local stakeholders, key land cover types that matter for both science and real-world decisions were defined. Under this framework, we adopted cutting-edge AI techniques to make the model strong and adaptable, even in complex landscapes.

Early results are promising: the model is over 90% accurate in identifying key land cover types, including vegetation types that look similar but function very differently ecologically. This means better tools for tracking ecosystem changes, understanding climate impacts, and informing local action. Ultimately, GRADIENTES lays the groundwork for more inclusive, science-based monitoring across the Andes and beyond.