Bayer Crop Science has engaged in a multi-year collaboration with Sparkgeo Consulting to deliver an evolving set of spatial imagery search, discovery, and visualization capabilities built on top of open source geospatial software. Initial solutions integrated CKAN, Geoserver, and Geotrellis to pre-render custom tilesets for derived analytic outputs. This process proved difficult to scale with increasing ingest rates and led to standardizing imagery pipeline outputs on Cloud Optimized GeoTIFFs(COGs) with rio-tiler, pyproj, GDAL , Shapely, and Rasterio for processing to define dynamic rendering visualization products in a newly developed STAC-compliant catalog. The Sparkgeo team has written a custom Global Imagery Search tool for our corporate OpenLayers-enabled application framework which combines event-based per-scene visualization processing with STAC search results and TMS-to-COG range/column searches. The Global Imagery Search tool also allows client/application side dynamic color map rendering. This presentation will describe the evolution from tilesets to dynamic rendered tiles and the customizations within STAC-collections needed to achieve this. This presentation will describe the implementation challenges of scaling inputs and processing pre-rendered tilesets for application visualization and how the decision to re-direct to a COGs + STAC cloud implementation has met our scaling objectives. Authors and Affiliations – Martin Mendez-Costabel – Bayer Crop Science Paul Trudt – Bayer Crop Science Will Cadell – SparkGeo Consulting Dustin Sampson – Sparkgeo Consulting Joe Burkinshaw – Sparkgeo Consulting Angelo Arboleda – Sparkgeo Consulting Track – Use cases & applications Topic – Business powered by FOSS4G Level – 1 - Principiants. No required specific knowledge is needed. Language of the Presentation – English |