Machine Learning and Visual Analytics for Environmental Science
Environmental science has entered the age of Big Data. Between elaborate sensor networks (both on satellites and at the Earth’s surface) and complex simulations, modern data sets sometimes contain billions of points. Each point associates many coordinates (three spatial, one temporal, plus simulation or policy parameters) with many values (temperature, pressure, moisture, density of plant and animal life, etc.). Using cutting-edge techniques from machine learning (e.g., deep convolutional neural networks) and data visualization, this project builds tools to help environmental scientists analyze these rich data sets and communicate their findings to community members and policymakers.