Cloud-native science is not constrained by the limits of personal computers. Cloud-native science means anyone could run the analysis, given access to the same cloud environment.
This panel gathers perspectives from leaders in cloud computing and geospatial data processing standards and tools. Panelists will give 5-7 minute talks on their areas of experience in using the cloud to support Earth science. During the talks, attendees will be encouraged to post their questions asynchronously to a google document or slido. Once lightning talks have completed, the most upvoted questions will be answered by the speakers.
Why cloud?
- Paige Martin will present "How cloud computing can enable open science": The COESSING program is a real-world example of how cloud-native has not just opened the door to unprecedented scales of science but also opened doors to participation. West African scientists who don’t normally have access to large datasets or powerful computing resources, but could greatly benefit from cloud-native science. However, there are current shortcomings, including training, the need for analysis-ready data, transparent cost structures, and avoiding cloud “silos” where each data provider has their own cloud interface to access their own data.
Create and manage cloud archives
- Brianna Pagán will present "Landscape on cloud-native formats and methods"
- Peter Marsh will present "Accessing NetCDF and GRIB file collections as cloud-native virtual datasets using Kerchunk"
- Vincent Sarago will present "TiTiler + TiTiler/pgSTAC"
- Kyle Barron will present "Cloud-Native Vector Formats"
- Ryan Abernathey will present "Arraylake - A data lake for multi-dimensional arrays"
How to use cloud archives
- Matthias Mohr will present "openEO: Interoperable Geoprocessing in the Cloud": openEO is an API specification for Earth Observation data cubes that supports discovery, data extraction, processing and viewing in an interoperable, federated and reproducible manner. Both the standard and most of its implementations are Open Source projects. Backend implementations exist for geopyspark/Geotrellis, dask/xarray, GRASS GIS, Rasdaman, Google Earth Engine and more. Clients are available for JavaScript, R, Python, and web browsers. This lightning talk will give a brief overview over the project, the API and process specification and implementations.
- Sam Roy will present "The GEE community catalog, the motivation, stories, and methods": The awesome-gee-community-catalog consists of community-sourced geospatial datasets made available for use by the larger Google Earth Engine community and shared publicly as Earth Engine assets. The project was started with the idea that many research datasets are often unavailable for direct use and require preprocessing before use. This catalog lives and serves alongside the Google Earth Engine data catalog and houses datasets that are often requested by the community. Come find out more about how you can contribute and get your datasets features and share and build the catalog together.
- Anne Fouilloux will present "RoHub: Research Object Hub Manage and preserve your research work, make it available and discover new knowledge."
Recording
Notes & More