Cloud-Native Geospatial Forum (CNG)

Registration for the first in-person Cloud-Native Geospatial Forum (CNG) Conference is now open! Join us on 30 Apr to 2 May 2025 at Snowbird in Utah. ▪️ Take advantage of our Super Early Bird period, available now through 31 Dec 2024 and save on your registration 📌 Link to register in the first comment!

How it became so easy to stream data from GeoParquet datasets 1.1 derived from Overture Maps, Source Cooperative and Four Square for a specified bbox ! Now, you can retrieve geospatial data for your POI in #GeoParquet format using DuckDB through GeoParquet Downloader #QGIS Plugin. thanks Chris Holmes for this amazing plugin. #geoparquet #cng #duckdb #geospatialdata #overturemaps #sourcecooperative #foursquare

If you’re working with geospatial data, chances are you’ve come across the name, Kyle Barron. As a Cloud Engineer at Development Seed, Kyle is one of those data engineers that are fundamentally changing how we work with large #geospatial datasets on the internet. In this interview, he shares his thoughts on GeoArrow and GeoParquet—two tools that "speed up handling larger amounts of geospatial data." Read the full interview on the #cloudnativegeo blog site.

New JS library to read #geoparquet: > GeoParquet provides a pure JavaScript workflow to read and convert GeoParquet files into GeoJSON. Under the hood, it uses the hyparquet library for efficient in-browser parquet parsing, enabling minimal overhead and fast loading. By the same author as hyparquet Kenny Daniel The repo: https://lnkd.in/evqQFeRY An Observable notebook: https://lnkd.in/eCkU-h2x

Can 2025 be the year when cloud computing transforms from powerful but complex to powerfully simple? I'm excited about making our efforts into cloud technologies truly accessible to every organization and their developers. We've already seen some game-changing approaches to making complex tech more human-friendly: The Project Jupyter ecosystem started with notebooks and grew into full-blown cloud environments (2i2c) that just... work. No more "but it works on my machine" or spending hours configuring environments - just science-ready compute when you need it. 😅 Platforms like Microsoft Planetary Computer and Google Earth Engine (GEE) have continued to show how analyzing global satellite data could be as easy as using desktop GIS. And now we're seeing game-changers like Microsoft and NASA-IMPACT's Earth Co-pilot (https://lnkd.in/gPGYJN9d) giving you an AI assistant in your map interface that not only understands your Earth science questions but also knows its way around massive satellite datasets. This is a future for geospatial in the cloud that I'm excited about. Making this future real needs two things: 1. UX & Frontend wizards who can make the complex feel simple (PS: We're hiring! https://lnkd.in/gQX25_fk). 2. Champions in organizations who can help teams embrace these tools & workflows. I know that lots of smart people are working towards this already! What are some other 2024 wins for adopting the cloud with increased accessibility and usability? 📷 : Borrowed this Development Seed pic. It's actually perfect because it represents our team members across earth data, AI, and application development, all talking together about realizing this very future that I imagine.

Thank you to Dr. Julia Wagemann for the chance to chat on one of my favorite podcasts #ScenefromAbove! I wasn't able to give my planned talk at AGU but this podcast is an extended reflection talking about the need for federated compute resources, discussing what could be the correct level of abstraction for data providers to distribute satellite data and the not-so-sexy but much needed work on metadata standards and their implementation holding our industry back. I also give some shoutouts to efforts like virtualizarr and folks at 2i2c, Openscapes and NASA TOPS. You can listen here: https://lnkd.in/enpzPXzZ

🚲 If you want to really learn how to work with data, you need to work with messy data. This post from Zhengpei Luo does a great job of showing this with the Citi Bike Dataset which is...messy. Here's why: 🤐 The data is stored in zipped directories (by year) or zipped CSVs by month (but just for 2015 and forward). 😡 The dataset changes schemas 3 times and in the final update the point is tied to the exact pickup location, not the station location. Plus there is a lot of data: 53,083,660 total trips. But here is the data for you to check out! This is just a sample of trips in June 2024 that I processed in Wherobots! #gis #moderngis #geospatial #spatialanalytics #spatialsql #duckdb #wherobots

We are grateful for Splunk's support, which makes it possible for us to provide more free educational resources, offer free CNG membership, and grant event scholarships to community members who are unable to pay.

Was a pleasure being interviewed for the Cloud Native Geo blog! https://lnkd.in/eCbxWFcn

A universal map URL: does something like this exist? (edit: maybe this could be called "geocode anything", use any input pair with the platform to output a map view on any other platform) You've probably used TinyURL or similar before. What might be an interesting location sharing tool is some kind of base URL, let's say map.sh (it seems abandoned). Then in the interface, you can choose a drop-down that switches which map platform you want to primarily use, it could be Google Maps, Apple, Qwant, OSM, etc. Then you choose either a map pin (lon,lat) or a POI. Then click to generate link. When the person you send the link to (map.sh/xhya6dha6 or some hash) opens it, they see a button to open it in various map apps, with one being green as the primary. If they open in Google Maps and you shared in Google Maps with the Natural History Museum POI selected, it opens that POI. If you choose another platform it converts to a lon,lat,zoom to still display the pin, or maybe even carries over the place name to find if same or similar exists on the next map. You could even make the URL easy to auto generate as like https://lnkd.in/etcy88vg or map.sh?map=qwant&lonLat=42.01345,12.54731. And instead of map you could even load in a satellite tile, like bought from SkyFi, or a custom map URL from something like MapTiler or Stadia Maps, or a static map with info overlaid, like the route of a taxi on its way, without having to deploy it permanently. Or maybe it's enough as maps are to just have a lonLat and choose your app, but I think about this sometimes. #gischat