DeepSensor Great Lakes
The Great Lakes, as the planet’s largest freshwater reservoir, not only provide drinking water for over 38 million people, but also serve as a vital resource for irrigation, shipping, hydroelectric power generation, and recreation. Because of their importance, observing and monitoring the Great Lakes is a critical activity. Due to the limited resources available for this much-needed observing and monitoring, it is crucial to make the most efficient use of available observing platforms (e.g. buoys, research vessels).
I am working with ressearchers at the Cooperative Institute for Great Lakes Research to develop and use an open-source package, DeepSensor to answer - “where should the next generation of temperature measurement sensors be placed in order to most efficiently improve our quantitative understanding of Great Lakes surface temperature variability?”.
DeepSensor is a package for probabilistically modeling environmental data with neural processes, to characterize Great Lakes surface temperature and to make informed suggestions for future temperature sensor locations.
Contributions so far: * Built experimentation pipeline to train DeepSensor on 11 years of Great Lakes geospatial surface temperature data. * Added functionality to include ice concentration data as additional context. * Optimized training process with efficient data chunking. * Generated predictions for surface temperatures and used meta-learning techniques to identify new sensor placements.