I forked this and put plate tectonic velocity vectors (in an earth-centered, earth-fixed coordinate system) in place of wind vectors but didn't have the .js knowledge or time to fix all the annotations and menu stuff. I'm hoping to get to that this fall (but I've been saying that for a year or two...):
The velocities are in mm/yr. 68 mph plate tectonics would definitely be terrifying; that's an order of magnitude faster than faults move during an earthquake.
There are two phenomena that look cyclonic.
The larger ones have to do with motion on a sphere: Euler showed that any velocity vector on a sphere can be represented as a rotation about a point (called an Euler pole); tectonic plates move basically rigidly, so the whole things rotate around a point. When that point is very close to (or within) the plate, the rotation is quite visible. You can see this in the Antarctic plate, halfway between Antarctica and Madagascar.
The second is an artifact of the vector visualization that Beccario used. The atmosphere is a continuum, so particles move smoothly between one location and another (more or less). But tectonic plates have rigid boundaries, and dive under each other or slide rigidly past one another. But the visualization algorithm doesn't know about these boundaries, so it creates particle paths that cross the boundaries and appear to swirl around. This probably happens to some degree in the mantle below the plates, but not really at the surface, with minor exceptions that I won't get into now unless people really want to nerd out on microplate rotation.
I'm not sure about the data gap at the poles. That's probably a real gap in the dataset I used; I haven't looked into it.
The point where velocity field vanishes seems to be a good example of the hairy ball theorem; even if the motion on the surface is not a rotation, if the field is continuous there has to be at least one point on the surface where velocity is zero.
I used this site to check wind conditions during the 2015 Tianjin explosions. I was 90 miles away (Beijing) and was considering leaving the country had wind been blowing in the wrong direction. At the time the risk of the hazardous chemicals released by the explosions was completely unknown, so wind conditions were the only objective piece of information I could base my decision on. Fortunately the wind was blowing north/east.
Right now is especially fascinating because the Euro and GFS models have recently begun to agree on the storm re-forming into a hurricane in a couple of days off the New Jersey coast. It's a fascinating time to watch wind patterns in the US right now!
The interface on windyty.com is better. It does not have the problem of filling up your browser's history with every move you make. I would like to be able to see the color scale key with out also having to open the full control panel.
Every new view should get a URL. That makes every view linkable. People who do not do this are actually violating the spirit of the HTML/URL specs. The fact that you see your browser history as polluted is an implementation detail that could be rectified by simply collapsing all consecutive URL's from the same domain over N url's into a hierarchical dropdown.
It can also be (sort of) rectified through judicious use of history.pushState vs. replaceState. The latter just replaces the URL in your URL bar but doesn't add an entry to your history - so every new view can get a new URL without polluting history.
I say "sort of" because everyone has their own idea about which things should be pushed vs. replaced. One person's "history pollution" is another's useful tool. But in this case it seems pretty clear that simply moving the map shouldn't push to history.
Thanks for sharing this. This is so cool for kitesurfing and sailing. I didn't know any of these services and windyty looks really nice. Specially I like that you can zoom-in much more than in OP's globe....
If you're interested in playing with this visualization technique (particle advection in vector fields) outside the context of geospatial applications - I built this tool called "Vector Toy" which allows you to visualize any given (2D) vector field this way: http://dandelany.github.io/vector-toy/ - note that the functions on the right side are editable.
If you use 500mbar visualizations, you can visualize the "reverse" weather pattern from last summer, with the wonky jet stream, high pressure baking the Pacific Northwest, and cool weather in the midwest and on the East Coast: https://earth.nullschool.net/#2015/07/03/1800Z/wind/isobaric...
Conversely, you can see how the "heat dome" of this summer smothered almost the entirety of the contiguous United States, except for the West Coast in general and the Pacific Northwest in particular: https://earth.nullschool.net/#2016/07/22/1500Z/wind/isobaric...
It doesn't reload data - it just stops the particle system advection and restarts it when you're done moving.
And yes, this is pretty necessary (at least it's quite hard to work around). To maintain the same view during viewport movement, it would have to keep track of all the particle trails in lat/long space and re-project them all to the new view. But this is implemented in canvas, so the trails (ie. particle histories) are likely not stored in their entirety - more likely, the system keeps track of only the particle's current position, and the trails are accomplished by drawing new line segments over the old on each frame. Yes, it would be technically possible to keep track of each particle's full history, and to re-project them all during interaction, but it would be pretty slow.
(I didn't make this, but I've written a very similar particle system in canvas and ran into similar issues)
I imagine you could accomplish the trails with some triangle strips in WebGL and let the GPU do the heavy lifting. I wonder if that would perform any better.
The location of this data point on the earth's surface is 9.8 degrees North Latitude, 149.27 degrees West longitude[0]. Wind direction is 360 azimuth degrees(meaning it's coming from straight north at that point sampled[1]) at a speed of 88 km/h.
this is amazing! so trippy. You can really track the two storms (US East Coast and right of Hawaii). The eye of the storm is relatively calm at 0-6 km/hr.
http://earth-analysis.com/vels/public/