Date:
Location:
Talk Title: A pragmatists' guide to machine learning for Earth observation imagery
Abstract: Our planet is constantly changing as a result of ecological and socioeconomic interactions across a variety of spatial and temporal scales. A growing number of Earth observation datasets provide regularly repeated synoptic measurements that enable us to map and understand landscapes as the dynamic systems they are. However, extracting meaningful information from petabyte-scale archives of multi-spectral, multi-temporal imagery requires a sound working knowledge of sensor characteristics, standard pre-processing workflows, and other domain-specific expertise. This talk will examine some of the unique challenges of applying modern machine learning techniques to Earth observation imagery with examples and lessons learned from ongoing research efforts.
Bio: Dr. Valerie Pasquarella is an Researcher at Harvard Forest, a department of Harvard University, where she is a Co-Investigator with the Harvard Forest Long Term Ecological Research (LTER) program, and she is currently a Visiting Researcher with Google's Geo Sustainability team. Working at the intersection of remote sensing and ecology, Dr. Pasquarella uses time series of satellite imagery to support improved mapping and monitoring of landscape patterns and processes. Her research interests include near-real-time and retrospective change detection, deep learning approaches for image understanding and feature extraction, and making Earth observation datasets more useful and usable for the broader scientific community. She holds a BA in Environmental Science, an MA in Environmental Remote Sensing & GIS, and a PhD in Geography from Boston University. She also served as a postdoctoral fellow with the Northeast Climate Adaptation Science Center at the University of Massachusetts Amherst from 2016-2018 and has held a Research Faculty appointment at Boston University since 2018.
If you would like to join our Seminar Series google group and receive a reminder about this event and upcoming seminars, please register here.