EarthCube Project Uses Deep Learning for Mapping and Classifying Sea Ice
Sea ice, an important component of the climate system and a key indicator of climate change, is spatiotemporally dynamic, exhibiting a variety of evolving ice types that need classification for scientific analysis or operational planning. The mapping of sea ice at high spatial and temporal resolutions remains a scientific challenge; however, with the increasing availability of high-resolution remote sensing products, such as synthetic-aperture RADAR and LiDAR, there are now opportunities for tackling this challenge, which is just what EarthCube-funded Harmonized Earth aims to do.
Harmonized Earth is a collaborative research and capabilities development effort for creating algorithms, models, software systems and cyberinfrastructure for harmonizing heterogeneous big data products (including satellite imagery and in situ observations) in a cloud environment for various downstream tasks. The technologies developed are expected to be extendable to a variety of applications, but for this project, the focus is classification and mapping of sea ice.
“Harmonized Earth, a collaborative research and capabilities development effort for creating algorithms, models, software systems, and cyber-infrastructure for harmonizing heterogeneous big data products, includes satellite imagery and in situ observations in a cloud environment for various downstream tasks,” explained Principal Investigator Morteza Karimzadeh, an assistant geography professor at the University of Colorado (UC) at Boulder. “The technologies developed are expected to be extendable to a variety of applications, but for this project, the focus will be on classification and mapping of sea ice.”
Karimzadeh has been working with collaborators at the National Snow and Ice Data Center and UC Denver’s Department of Computer Science to bridge the gap between data science and geoscience in order to harness large heterogeneous data for sea ice mapping. The project brings geospatial data scientists, geoscientists and computer scientists together, using state-of-the-art machine learning coupled with satellite observations for seamless data fusion, machine learning and analysis.
Harmonized Earth is a collaboration between the Principal Investigator Dr. Morteza Karimzadeh (CU Boulder Geography), Andrew Barrett (NSIDC), Walt Meir (NSIDC), Siri Jodha Khalsa (NSIDC) and Farnoush Banaei-Kashani (CU Denver Computer Science).
Harmonized Earth is funded by National Science Foundation awards 2026962 and 2026865.
EarthCube is a community-driven activity sponsored by the National Science Foundation to transform research in the academic geosciences community. EarthCube aims to create a well-connected environment to share data and knowledge in an open, transparent, and inclusive manner, thus accelerating our ability to better understand and predict the Earth’s systems. EarthCube membership is free and open to anyone in the Geosciences, as well as those building platforms to serve the Earth Sciences. The EarthCube Office is led by the San Diego Supercomputer Center (SDSC) on the UC San Diego campus.
Kimberly Mann Bruch, San Diego Supercomputer Center Communications, firstname.lastname@example.org
Lynne Schreiber, San Diego Supercomputer Center EarthCube Office, email@example.com
Harmonized Earth: https://www.youtube.com/watch?v=Wy-r5O6teLs
San Diego Supercomputer Center: https://www.sdsc.edu/
UC San Diego: https://ucsd.edu/
National Science Foundation: https://www.nsf.gov/