Skip to content

CNG (Cloud Native Geospatial)

Summary: Notes on Cloud Native Geospatial formats, standards, and tools such as COG, GeoParquet, and Zarr. Last updated: 2026-08-20

2026

  • Cloud Native SDI - Portolan: Cloud Native SDI — presentation exploring how cloud-native geospatial formats (GeoParquet, COG, PMTiles, STAC) can transform Spatial Data Infrastructures, introducing Portolan, an open-source framework for building serverless, AI-first SDIs, demonstrated via a live Dutch government data portal requiring zero servers. Related: Web_Mapping, STAC, AI. Keywords: cloud-native, GeoParquet, COG, PMTiles, STAC, SDI, Portolan, serverless

  • dynamical.org: A research lab providing a public catalog of cloud-optimized weather and climate datasets, helping researchers, forecasters, and developers access, understand, and act on environmental data. Related: Climate, Data. Keywords: weather, climate, data, cloud-optimized, open data, forecast

  • EO Platforms, Science Platforms, and the Apartment Between Binder and Earth Engine: Blog post on the current state of Earth Observation platforms, arguing they exist on a spectrum from managed services (Google Earth Engine) to bare infrastructure (Kubernetes), with "apartment model" namespaces offering identity, storage, and compute as an optimal middle ground. Related: Remote_Sensing, GEE. Keywords: Earth Observation, platforms, Google Earth Engine, Kubernetes, cloud-native, infrastructure

  • spatial-access-measures: Web map using DuckDB-WASM and GeoParquet for in-browser SQL queries and filtering of Statistics Canada spatial access measures (transit, cycling, and walking accessibility to jobs, schools, and services). No server backend required — everything runs in the browser. Demo Related: DuckDB, Web_Mapping, Tools. Keywords: DuckDB-WASM, GeoParquet, deck.gl, web mapping, in-browser SQL, accessibility, Canada, cloud-native

  • deck.gl-raster: Client-side visualization library for large raster datasets like COGs and Zarr arrays directly in the browser using WebGL2 through deck.gl, with GPU-accelerated processing for color mapping and reprojection. Related: Web_Mapping, Remote_Sensing. Keywords: deck.gl, COG, Zarr, WebGL, raster, visualization, cloud-native, JavaScript

  • S2Mosaic: Python package for creating cloud-free mosaics from Sentinel-2 imagery, with flexible scene selection, multiple compositing methods, and state-of-the-art cloud masking via OmniCloudMask. Related: Remote_Sensing, Python, STAC. Keywords: Sentinel-2, cloud masking, mosaicking, Python, remote sensing, STAC, geospatial

  • Freestiler: R and Python tool for creating PMTiles vector tilesets from spatial data, including from DuckDB SQL queries. Uses a Rust tiling engine for in-process performance, producing single .pmtiles files compatible with Mapbox and MapLibre formats. Related: Web_Mapping, DuckDB, Python. Keywords: PMTiles, DuckDB, R, Python, vector tiles, cloud-native, MapLibre, geospatial

  • Yosegi: Python tool that generates pyramid GeoParquet files optimized for efficiently streaming large geospatial datasets. Uses density-based thinning, Sort-Tile-Recursive spatial packing, and GeoParquet 1.1 bbox covering metadata for fast overview queries and row-group pruning. Related: Python, Data. Keywords: GeoParquet, Python, pyramid, cloud-native, spatial data, optimization, streaming