cloud-native-geospatial

GeoZarr multiscale pyramids for planetary-scale rasters: a lazy, windowed and resumable writer of zarr v3 stores with sharding, for any xarray object with a CRS.

Kentucky's KyFromAbove program provides statewide 2ft DEMs, 3-inch orthoimagery, 3-inch oblique imagery, and COPC LiDAR point clouds — all publicly available on S3 with a STAC API for discovery. abovepy gives you Pythonic access to all of it, plus server-side terrain analysis via TiTiler. No credentials required.

FlatGeobuf Overviews (FGBO): a scale-optimized, fully fgb-compatible vector format — precomputed simplification, overview levels, and segments for fast on-demand MVT tiles from one static file. Rust encoder/decoder, CLI, tile server, and wasm browser reader.

Cloud-native geospatial related content/activities at the 2025 ESA Living Planet Symposium

FaultLine is a cloud-native geospatial change engine for comparing, validating, and tracking spatial dataset evolution across modern formats like GeoParquet, COG, and PMTiles.

Compares VKG approach against PostgreSQL + REST API for cloud-native geospatial data integration. Federates Sentinel-2 COGs (raster) and Overture Maps GeoParquet (vector) via DuckDB FDW, PostGIS, and PL/Python. Evaluates two raster integration strategies, and SPARQL (Ontop VKG) vs REST API (pg_featureserv) trade-offs. Fully Dockerised.

Validate whether a GeoTIFF is a proper Cloud-Optimized GeoTIFF and explain exactly why it is not: structural checks, HTTP range-request cost, and a derived gdal_translate fix.

A Cloud Native Geospatial study: dynamic spatial queries served entirely from the browser via DuckDB-WASM over Overture GeoParquet.

Curated resources for geospatial AutoML, algorithm discovery, agents, datasets and benchmarks.

OGC Building Blocks for the Portolan specification.

An analytical MCP server for SpatioTemporal Asset Catalogs. Safely orchestrates AI-driven spatial workflows and pushes compute to the edge using Rust and DuckDB.

See your distortion. A visual-first geospatial diagnostics engine — projection x-ray, cartographic linting, spatial diffing, and autofix. Rust core, Python bindings, browser-ready via WebAssembly.

Named analysis functions over FAO FERSPAS, each served as XYZ tiles with a time axis. One file per function; nothing pre-built.

STAC dataset intelligence: find it, verify it, plan it, replay it.

Turn a georeferenced raster time series into an interactive dataset anyone can open with a link, with the stored values intact. Static files, no tile server.