Show HN: Near-realtime cloud maps for weather visualisation, 3D models and more | Hacker News Reader

Show HN: Near-realtime cloud maps for weather visualisation, 3D models and more

github.com

68 pointsmatteason16 comments

This is a service I've built to provide equirectangular cloud maps of the whole Earth, updated every 3 hours, for free.

It's based on infrared and visible light data from EUMETSAT, processed to enhance the clouds and try to reduce the impact of surface temperatures on the final image.

I built it because it's surprisingly hard to find these kinds of images in a format that's useful to developers and hobbyists. I was building a 3D ISS tracker and wanted to include live clouds but couldn't find anything good, so I got sidetracked, and here we are.

Open articleView on HN

Man, that is so damn cool! Thanks!

Don't have any particular use case for this, myself, but its open accessibility (a simple url) makes it a drop in solution for all kinds of environments, including Unreal and Blender. And just based on what those two apps can do with it, I bet there are a bunch of others that can do some other really awesome stuff with it, too!

Thank you! Yeah I've tried to make it as no-nonsense as possible. I'm planning to add some other handy bits around it (eg a txt/JSON file giving the last update time) that might be useful for some types of project.

This looks cool! i want to build a persistence of vision globe (sealed in partial vacuum under a bell jar) with live cloud cover, and was looking for a way to get cloud data. is the source code for the generation of these images available?

Oh amazing! I've really wanted to make some kind of spherical desktop display so I'd love to hear how you get on.

You can just use the image URLs directly if you want to, but if you want to do more processing on the source images the code is all in index.js in the linked repo. It's just a Node script. I've tried to make it clear what's going on in the comments so it should be pretty straightforward to port to other languages if JS isn't your thing.

Came here for AWS - still left happy discovering it was about real clouds

Looks great!

Just one note: there is a very faint shimmer of a white line on the left and right edges of this image which thus becomes more visible when wrapping around a model

https://camo.githubusercontent.com/1a35eff2061fe218236f872f9...

Thanks! Yes you're right, it does become a lot more pronounced when it's used on a sphere. I'll find a nice way to fill in the gaps that doesn't just make it more obvious

Awesome! Thanks for putting this together.

Is there any way to get non-processed image data? Without the nighttime adjustments?

You can grab the source data from EUMETSAT directly - the URLs are all listed in https://github.com/matteason/live-cloud-maps/blob/3be662483c...

The products used are:

"Geostationary Ring IR10.8 μm Image - Multimission" - https://navigator.eumetsat.int/product/EO:EUM:DAT:0330

"Geostationary Ring Dust RGB - Multimission" - https://navigator.eumetsat.int/product/EO:EUM:DAT:0334

"Geostationary Ring Natural Colour RGB - Multimission" https://navigator.eumetsat.int/product/EO:EUM:DAT:0336

Looks Fantastic! I love stuff like this, and am jealous of Planet employees ;-)

Wonder how it compares with the GOES raw data, and also I wonder if a hybrid viewer would have any aesthetic benefits.

Do share your "live 3d ISS viewer" when done. I can imagine myself getting a smile or two from such a thing. :)

You can see the source IR image (which incorporates GOES data) here: https://navigator.eumetsat.int/product/EO:EUM:DAT:0330

It took quite a while to work out how to process it to get a level of realism I was happy with. IR is tricky because the same level of grey could either be warm cloud or cold land, so just discarding anything below a certain level to try and get rid of the land also loses a lot of cloud detail. I've combined that IR map with another false-colour image which lets me recover some of that detail, then I'm also extracting clouds from a true-colour image which restores more detail for areas in daylight.

> It took quite a while to work out how to process it to get a level of realism I was happy with.

YUP!

I haven't made a cloud viewer, but I once fell into "the deep end" regarding false color and astronomy photos. Next thing you know you're deep in MATLAB playing with histogram equalization...

Going from the raw data that advanced sensors output to a beautiful picture is NOT always obvious, there's a lot of enjoyable applied math and programming in between. =)

I hate to admit my first thought was "Cool something to help visualize my AWS account."

I know, I'm sorry! It's very hard to name something like this without including "no, not _that_ kind of cloud"

Cloud cover map, maybe?

I would suggest "cloud (the water kind)" but I hate it.

Reply on news.ycombinator.com