Résumé :

Exploration of remote areas such as Earth’s seafloor or Icy Moons in our Solar system often involves limited ‘time on target’ and no ability to return to areas of interest because of environmental, operational, and financial constraints. One way of maximizing scientific return is to achieve more knowledgeable decision-making at the time of exploration. This can be achieved by creating a framework before exploration of what to expect using modeling tools and then comparing expectations with observations in real-time. As part of the NASA-funded project SUBSEA, I developed this approach using geochemical modeling for the exploration of the Seacliff hydrothermal vent field, seen as an Earth’s analog of Saturn’s icy moon, Enceladus. In this presentation, I will show you how forward geochemical modeling allows (1) identification of the most informative parameter to measure on the field and (2) improvement of result-informed decision making during scientific exploration. Building on this work, I would like to initiate discussions on how the multiplication of a priori mechanistic modeling could be coupled to machine learning technics to refine our predictive capabilities of remote, unexplored areas on Earth and beyond.