Show HN: Realtime Geostationary Earth Imagery
bluemarble.nitk.inEarth from space, as of ~30 minutes ago.
This visualization uses imagery provided by GOES-R, a geostationary earth-observing satellite, to generate near-realtime true-color images of the earth at (fairly) high resolution. (There's also animated video, top-left.)
NOAA provides raw data from the GOES satellites available in AWS and Google Cloud, if you want to play with/remix it. The source data is incredibly high resolution - 100 or 400 megapixels, depending on layer. Significantally better than the imagery weather.gov exposes. Everything's provided as netCDF, but it takes some time to find the layer you're looking for.
- https://www.noaa.gov/organization/information-technology/lis...
- https://storage.cloud.google.com/gcp-public-data-goes-16/ (ABI-L2-CMIPF is the product I use)
The source code is on GitHub. It's built with a Python backend and some pretty simple javascript to handle auto-refreshing.Should be fairly simple to hack around with.
- https://github.com/bnitkin/goesr-video
Hope y'all think space imaging is as cool as I do!
Nice work Tim! What tech stack did you use to build this?
Not much of one :)
The backend is all Python - netcdf library for reading; Pillow for image manipulation. There's a second Python script that regenerates the videos daily, using FFMPEG. A simple cronjob restarts the services if anything crashes.
From there, rsync mirrors the files to a DigitalOcean VM running Apache.
The frontend is vanilla HTML/CSS/JS. No jQuery or other magic. The JS periodically changes the img href attribute to force an image reload, so if you leave the page up, it'll slowly change as the day passes.
Video page is similar - the javascript snippets load different .webm videos. (As an aside, cross-browser compatibility for embedded videos is challenging. webm had the best support when I built this, but it's far from universal.)