2nd GEO - AWS Earth Observation Cloud Credits Programme Webinar
News / 13 March 2019
Presentations and recording
In the second edition of the GEO - AWS Earth Observations Cloud Credits Programme Webinar series, hosts from NASA / CEOS, Sinergise, Development Seed and Radiant Earth Foundation will outline tools, software and services that help users leverage big Earth observation data and the AWS cloud computing environment.
The Group on Earth Observations (GEO) and Amazon Web Services (AWS) Earth Observation Cloud Credit Programme offers GEO Member agencies and research organizations from developing countries access to free cloud services to help with the hosting, processing and analysis of big data about the Earth to inform decisions for sustainable development.
Overview of Presentations
Brian Killough of the CEOS Systems Engineering Office at NASA will present the tools and services available to GEO-AWS participants for implementation of an open data cube in the Amazon cloud. Tools include installation modules, training materials, and sample application algorithms. The presentation will also discuss the service options for creation of data cubes using Landsat or Sentinel analysis-ready data.
Grega Milcinski of Sinergise will then present how the Sentinel Hub works around challenges of understanding various satellite imagery data formats, projections, bands, indices, etc., explaining how to easily and instantly access Sentinel, Landsat and MODIS data for visualisation, data processing and/or machine learning tasks, using a set of simple APIs. Special focus will be given to Python tools which make it possible to configure a data processing workflow in order to prepare the satellite data to the level required for machine learning frameworks such as TensorFlow, MXNet and others.
Hamed Alemohammad of Radiant Earth Foundation will present recent advancements in the SpatioTemporal Asset Catalog (STAC), an open source standard for managing geospatial data. Next, he will introduce MLHub Earth, a central repository of open source training data for machine learning applications that can be hosted on AWS. A demo of the MLHub Earth crowdsourcing toolbox for generating training data with a STAC compliant catalog will be presented. Participants will also learn how to use MLHub Earth API to find and use training datasets.
Ian Schuler and Aimee Barciauskas of Development Seed will share open source tools that they are developing with NASA for managing planetary scale data on the cloud, including Cumulus and sat-utils. They will discuss open source tools for applying machine learning, and will share suggestions on designing tools to provide insights from satellite data to decision makers.