Towards an automatic generalized machine learning approach to map lava flows

Abstract

Volcano-related resurfacing processes can be monitored by complementary using radar and optical sensors. Combining both data sources with machine learning (ML) approaches is fundamental to automatically extract volcano-related features. Here, a generalized ML approach is developed in Google Earth Engine (GEE) to map lava flows in both near-real time (NRT) and no-time critical (NTC) time scales. A first attempt towards a generalized classification to automatically map new lava flows is proposed.