COMEONBOARDPSG! How to empower PRISMA Second Generation high temperature events identification with on board deep learning approach.

The ComeOnBoardPSG!Autonomous computing with on-board machine learning for optimized acquisition and real-time analysis of hyperspectral data on PRISMA Second Generation – is a project funded by the Italian Space Agency (ASI) and it aims to enhance the capabilities of the upcoming Italian PRISMA Second Generation (PSG) hyperspectral constellation. The project is divided in three main scientific/technology tasks: 1) EC solutions to optimize data transmission; 2) Autonomous operations and decision-making and 3) Onboard storage and real-time onboard data processing for early alert systems In this study we focus on the third activity: the real-time onboard data processing for early alert systems which aims to provide timely and accurate warnings in relation to High Temperature Events (HTE) including wildfires. This implies a significant reduction of the latency time allowing to provide critical information. In the frame of the project, we advance preliminary results by expanding the training dataset and improve model robustness across varying wildfires characteristics.