Environet: A Project Update

The goal of the EnviroNet project is to present interesting environmental science problems through a repository of multiple 'ImageNet' analog labelled datasets to engage the wider artificial intelligence community and help track progress of machine learning research in tackling planetary scale geoscientific problems such as climate change through organised challenges. This EnviroNet project update will overview labelled datasets under consideration at different stages of development, from global climate models, reanalysis, multi-spectral satellite data, automated sensor networks on the ground to crowdsourced photos, that fit the EnviroNet framework. Interesting machine learning applications from localisation and detection, forecasting extremes, predicting future video frames, tracking Earth surface and atmospheric changes, to visualizing future climate extremes with generative models, being developed to address problems across the entire Earth system will be presented. In addition to a scientific overview of EnviroNet datasets and machine learning challenges, progress on organisation of EnviroNet, from hosting, labeling campaigns, to opportunities for sponsorship of competitions, conference or workshop sessions, will be discussed further.