FOSS4G 2023

The SpatioTemporal Asset Catalog (STAC) specifications are a flexible language for describing geospatial information across domains and for a variety of use cases. This talk will present the current state of the specifications, which includes the core STAC specification and the API specification built on top of OGC APIs. While the core specification has been stable for roughly two years and doesn't need a lot of updates, the API specification got numerous updates and is finally close to a stable release. This presentation digs into additions to STAC extensions and the latest community developments. We survey the updates to the open-source STAC ecosystem, which includes software written in Python, Node.js, and more. Finally, let's also look into the near future.

One decade ago, we saw the launch of the first earth observation cubesats by Planet Labs. In the years since we have seen hundreds of satellites launched, and dozens of startup companies launching taskable satellites. While this has led to incredible opportunities to leverage multiple sensors and sensor modalities, the massive increase of data has also created challenges in data management, discovery, and usage. The community driven SpatioTemporal Asset Catalog (STAC) specification was an important step forward in exposing data to users in a standard way that enables cloud-native workflows and has been successful across government and industry.

The process of actually tasking satellites, however, is still very much non-standard; each data provider exposes a unique API, if at all. Some data aggregators have created a single tasking API that proxies and translates to multiple data provider APIs, but this is still non-standard, and proprietary.

Element 84 has been leading an effort to create a community standard API around how users order future data and how providers respond to those requests. Working with government groups, commercial satellite operators, and data integrators, we have hosted working sprints to develop a specification and open-source tooling demonstrating the power of a tasking API specification.

This talk will cover the current status of the community tasking API specification, future plans, and a demonstration of how to use the API to order data.

“Cloud-Native Geospatial” is a new paradigm for performing efficient data access and compute the cloud in an interoperable way in order to achieve scalable and repeatable analysis of geospatial data. The last few years have seen major developments in open standards and open software that make this possible, supporting full end to end interoperable workflows on remote sensing data, starting from data discovery to publishing of derived products.

This talk will provide an overview of what Cloud-Native geospatial is and why it is important for building scalable architectures. It will cover the current state of the Spatio Temporal Asset Catalog (STAC) specifications, and the landscape of cloud-optimized file formats, for raster, vector, and point-cloud data formats (COG, GeoZarr, GeoParquet, COPC).

Earth-Search is a publicly available SpatioTemporal Asset Catalog (STAC) API providing an index for some of the public datasets available through the AWS Registry of Open Data (RODA) and has been shown to be a valuable resource for accessing the Sentinel-2 archive as Cloud-Optimized GeoTIFFs. A new version of Earth-Search is an update and enhancement of the Sentinel-2 metadata as well as new Collections of data available on AWS, including Landsat Collection 2, NAIP, and Sentinel-1.

This talk will include a summary of the STAC catalog, what STAC extensions are used and how the data is best accessed based on file formats. We will also dive into the datasets that are available through the API and will present the architecture for indexing including a discussion of data latency. We will provide resources and tutorials for how to get started with public geospatial datasets on AWS.