Cloud Native Geospatial Outreach Day

A gentle introduction to foundational geospatial data concepts, with a focus on raster data specifically. No prior knowledge is necessary, this talk will assume you are starting from scratch. Facilitated by Alex Leith, and will transition immediately to the next session (same zoom connection)

This talk will cover an introduction to Digital Earth Africa, including our software and data ecosystems, and will dive into a brief training course on doing data science in our Jupyter environment.

You'll learn how to view our Cloud Optimised GeoTIFFs on a web map, how to analyse Earth observation data in Jupyter, using the Open Data Cube, and about how you can use our Sentinel-2 analysis ready data directly in your applications.

Overview of the STAC Specification. Will give background on the effort, the motivations behind it, and dig into each of the 'sub-specs': Item, Catalog, Collection and STAC API. The session will then be an open forum to ask any question about STAC, to help get everyone up to speed.

-Sentinel Hub - overall intro (Grega Milcinski, Sentinel Hub) - 20 minutes

-COG@Sentinel Hub (Max Kampen, Sentinel Hub) - 20 minutes, pre-recorded

-Q&A (Grega Milcinski, Sentinel Hub)

Welcome & Opening Remarks - Chris (Planet) + Bruno (Microsoft)

Data Labeling Contest Details - Hamed Alemohammad (Radiant Earth Foundation)

Cloud Native Geospatial Overview - Chris Holmes

Planetary Computer + AI for Earth - Bruno Sánchez-Andrade Nuño (Microsoft)

I will present a brief overview of Digital Earth Africa, including what we aim to do, why free and easy to use Earth observation data and information products are important, and how this continental scale effort is powered by STAC/COG.

Brief overview of why/how we use STAC at Arturo and the STAC related tools and catalogs we have built.

We will describe the AWS Open Data Program including current datasets available in cloud optimized formats.

Why Maxar chose to use STAC and how we have implemented it.

A recent STAC implementation that exposes and proxies NASA's CMR geospatial data.

This talk will give a brief introduction to zarr's functionality including compression, data storage and parallel computing. This talk will also demonstrate the integration of zarr with data discovery/cataloging tools such as intake.

In this talk we will present the current status and future plans of Sentinel-5P Level-2 and Level-3 datasets available in the AWS Open Data Registry: native data format are collected in Frankfurt region close to other Copernicus datasets to support the development of cross-satellite applications, while Cloud Optimised Geotiff products, including daily mosaics (Level-3) products, are generated to boost the exploitation of Ozone (O3), Nitrogen Dioxide (NO2), Sulphur Dioxide (SO2), Carbon Monoxide (CO), Methane (CH4), Formaldehyde (HCHO) and Aerosol products.

User interfaces to support data exploitation will be presented, including STAC and ADAM Explorer / adamapi.

I shall present DotNetStac library with a practical how to using small quick examples. I shall complete it's usage in a more complex project of a command line tool to read and navigate in STAC catalogs

Quick overview of COG file organization and how GDAL uses it, presentation of the new COG driver of GDAL 3.1

How the STAC specifications are being exploited in addition to OGC API Features interfaces in the ESA ESE-ERGO Project to allow access to ESA FedEO and EOCAT assets simultaneously accessible using OGC OpenSearch interfaces with Atom and GeoJSON responses.

This lightning talk will detail the capabilities of remote data exploration of potentially huge COG files using the geotiff.js library and the COG-Explorer Browser Application

How we have implemented STAC API to provide access to Sentinel Hub's catalogue of global archives of Sentinel-1,2,3,5p, Landsat-8, MODIS and other datasets, technical information on data and its storage. We will also discuss deviations from stac-spec due to implementation details.

Differences (and similarities) between terrestrial spatial data and planetary data, unique challenges to planetary data, how STAC is helping us meet or goals for cloud-based data release of derived data products.

OGC standards and practices must work in the Cloud. Extensive work in OGC's Innovation Program has been performed to assess suitability of existing standards or identify requirements for new standards for Cloud-Native operations. OGC recognizes that the geospatial community is evolving tailored specifications to work most effectively in the cloud and proposes means to endorse those specifications as part of the community standards baseline.

Along with the STAC sprint, Radiant Earth is hosting labeling competition--this talk will walk the audience through how to sign up, how to get started labeling, and how entrants will be judged.

Principles and patterns for querying imagery from a STAC API-compliant web service

In this beginner session we'll walk through creating a STAC of open data using PySTAC in a Jupyter notebook.

A beginner-friendly session (no geospatial or ML background needed) that walks through a common ML-for-mapping workflow (mapping building footprints from overhead imagery) with emphasis on what/why/how labeled data is important, defining key geospatial ML terms and concepts, and how training datasets are enhanced by using COG and STAC. As an example, we'll also introduce the Open Cities AI Challenge dataset of drone imagery (as COGs) and building footprint labels across 10+ African cities organized in a STAC.

Titiler is a modern cloud optimized geotiff tile server with lots of cool features. Arturo-stac-api is a STAC-compliant API with an OGC Tiles/Titiler integration to render map tiles from the catalog. Both projects are backed by FastAPI and part of a larger goal to make it easier to configure and deploy cloud native geospatial architecture.

A recently open-sourced project, Cirrus, is an AWS pipeline for scaling up processing and publishing of STAC metadata and assets. Use Cirrus to just publish STAC Items, or convert some of the assets to COGs while mirroring the others as is.

In this talk users will learn how to install the QGIS STAC Plugin, configure the plugin for a STAC API, browse collections within the APIs, and download assets from those APIs.

This intro session will walk people through getting started with Franklin in a more hands on and more thorough way than is currently available in the documentation.

Topics to cover will be:

- starting the Franklin service

- getting data into Franklin

- STAC transactions

- serving raster tiles

- validation and fixing items with transactions

- talking to Franklin with PySTAC

In this talk, we will go over the different tools you can use to validate STAC items

In this Lightning Talk, we will describe how and why we embraced STAC early on in the development. We will discuss some of the projects we were able to integrate STAC into and why we choose to go that route.

Earth Engine is the combination of a large data catalog with a planetary scale computation platform to analyze and visualize Earth data on Google's infrastructure. The system currently has 35PB of open Earth data in over 550 collections with many of those collections continually updating. Earth Engine’s mission is to expand our knowledge of the planet and use the system to power solutions to global challenges. Combining data and computation, scientists, researchers, and developers can detect changes, maps trends, and quantify differences across the planet. Users of the system must be able to find and use that massive data collection for Earth Engine to be effective. We have developed an internal catalog format based on ProtoBufs expressed as YAML. Each collection in the catalog contains basic metadata, a thumbnail view, and sample code to get users started working with the collection. While this format has worked well so far, we would like to switch to the SpatioTemporal Asset Catalog (STAC) format to enable several goals: 1) User contributions to the public catalog need catalog entries. 2) User contributed improvements to existing catalog entries. 3) Take advantage of existing STAC tools to help enable exploration of the Earth Engine catalog. 4) User owned and maintained (public or private) catalogs in cloud storage. 5) Enable more interaction between Earth Engine and systems that are outside of the core Google infrastructure.

A quick look into STAC Index, the list of STAC catalogs, APIs and tools that offers a free STAC Browser for all datasets

Cloud Native Geospatial has started on the raster world, but vector is as important. The cloud-first approach to vector these days come from the next generation of data warehouses. Google Bigquery, Snowflake, Azure Synapse and AWS Athena are some of the players on the space competing. The proposal for storage-computing separation, multitenancy and scalability will redefine how spatial data infrastructures get build on the future. CARTO is been applying this into its Data Observatory and in this talk we will talk about it.

This talk will be about Astraea's use of standards-based formats and interfaces, such as COG and STAC API, to allow fast and interoperable access to satellite imagery, allowing data scientists to more easily query, explore, and analyze large imagery collections.

I'll briefly describe the two stacks developed to ingest CBERS data into AWS: one responsible for data transfer and COG generation and the other responsible for indexing the content using STAC.Most of the developed software is open, I'll provide links and discuss the architecture.

Attendees will learn about this experience and the online resources available.

Adoption of Remote Sensing in subsaharan Africa, and the bottlenecks that is still slowing down it's full adoption

Building on our experience in data discovery in the context of spatial data infrastructures, we have learned that common standards-based dataset metadata is not suitable for all types of content, including specifically large scale collections of earth observation data. Using our open source Geoportal Server we have built an index that addresses these challenges with large collections of content. In this talk you will hear about these developments and see them in action.

A high-level overview of Radiant MLHub mission, and STAC catalog to develop a ML commons for Earth observation.

The future of geospatial data visualization is in the browser. WebGL enables fast client-side satellite imagery processing. But users are far from the source data, and direct access to large collections of Cloud Optimized GeoTIFFs (e.g. Landsat 8 on AWS) would incur unacceptable latency. With COGs, it's possible to quickly combine a mosaic of imagery on demand into tiles ready for consumption in the browser. I'll provide an overview of my recent open-source work to enable fast dynamic image tiling, and show examples of its potential in the browser.

Creating a STAC catalog out of radio-occultation data is an interesting and unique challenge. I want to discuss some of the successes and failures we've encountered in doing so, and I hope to provide an interesting and off-the-beaten path example of STAC in action.

This talk will go over cool stuff you can do with GeoRasterLayer, a plugin for easily visualizing Cloud Optimized GeoTIFFs on a Leaflet web map.

A common pattern for geospatial analysis goes like this: 1) search and discover data. 2) download… and wait. 3) open and analyze with your favorite programming language. What if you could skip the downloading and go straight to interactive analysis in Python? You can with “Intake-STAC”, a Python package designed to simplify geospatial data management! Tune in for quick example using Cloud-hosted public data.

Planet has developed an approach for working with raster data in the browser with OpenLayers. A brief demo of how it's used, how it works, and implementation details will be covered.

Introduction to the STAC Pointcloud Extension and the STAC-ification of the USGS 3DEP Lidar Dataset.

Why we chose to use STAC for all our infrastructure from day one at Near Space Labs. How we use a protobuf version of STAC and gRPC for our internal services and for our public API

NASA's Earth Observation Division supported the Covid-19 SpaceApps Challenge (https://covid19.spaceappschallenge.org/) through sharing data to AWS Public Datasets Program. Working with NASA stakeholders and science advisors, we produced Cloud-Optimized Geotiffs for MODIS Aerosol Optical Depth, Vegetation Indices and OMI NO2 products. To support ocean scientists and UW OceanHackweek, we worked in a similar fashion to produce MUR SST in Zarr.

These products enable users to explore trends in these environmental variables. During this talk, we share important factors that went into our process of delivering COGs and Zarr stores and what should be replicated or improved on in the future.

Presentation of resto server and rocket client ecosystem

This talk will provide an overview of pygeoapi and discuss recent STAC updates and enhancements.

The talk will give an intro to TileDB and its use as a cloud native universal data engine for geospatial data from point clouds to hyperspectral!

This session will introduce the MUR SST Dataset the Zarr data format. The MUR SST dataset is stored in Zarr on AWS PDS. Sea Surface Temperature is a key climate change variable and is being used to study how ocean temperature can impact other environmental factors such as biodiversity. This session will demo the dataset itself, as it was used during OceanHackweek to study marine heat wave variability. We will discuss the benefits of the Zarr storage format when compared with traditional file stores and cover how different Zarr chunk shapes can impact query performance.

Intro to open source tools for geospatial data analysis and visualization

Inake-STAC is an open source Python package for discovering, exploring, and loading spatio-temporal datasets. It facilitates working with geospatial data by providing default visualizations and functions for directly loading data into Python objects like Xarray Datasets or Pandas Dataframes. Using intake-STAC allows you to stop worrying about file paths and urls and instead focus on analysis. You'll also learn how Cloud-optimized Geotiffs (COGs), Xarray, and Dask work in a complimentary way to parallelize computations on raster data.

This session will dig deeper into how COG & STAC work. It will go into each of the 'sub-specs': Item, Catalog, Collection and STAC API. And it will provide a deeper dive on how COG's are set up. The session will then be an open forum to ask any question about STAC or COG, so if you still have questions at the end of the day come here and get them answered! It will be similar to Matthias's session at the beginning of the day.

I will provide a brief overview of Radiant MLHub catalog and datasets. Then present a live demo of how users can use our STAC API to search and access training datasets. I will use the LandCoverNet as an example, share the notebook in advance so participants can follow the demo live.