Celsius please. But besides that: Very nice, very clear and way better than supid sun and cloud icons with temperatures next to them over some vague areas.

I am proud of forecast.io for sticking with a 'normal' color palette here, rather than the non-standard purples they have chosen for radar; as a meteorologist the purple radar hurts my brain, but this map is completely cool.

Also there are definitely some quality control issues with the RTMA data, so you can bump into the garbage-in-garbage-out issue with it, but overall, this is a very nice start to something that could ultimately be quite useful.

This is pretty impressive, i wonder what the update interval and source is.

http://www.mapbox.com/labs/forecast/temperature/ I really like the interface, even the zoom buttons could be removed or hidden as you can scrool-zoom.

Make the temperature in Celsius please!

EDIT: There seem to be a problem with the algorithms, the largest zoom displays colder temperatures( look at the temperature map ).

Largest zoom,

http://i.imgur.com/GVt3iSy.jpg

Normal zoom, picture was resized by me

http://i.imgur.com/t1GgrOM.jpg

This isn't a map. This is a simulation. You made a simulation, and are outputting the results.

This teaches you nothing whatsoever except that your model has pretty colors.

Next thing someone is going to take these results, use them as input data for a new model, then send the results of that new model back as data for the first.

Hmm, there are some superlatives ("highest resolution") in here I'm not sure about. Lots of groups create maps like this for modeling purposes. One for sea surface temperature is:

http://ourocean.jpl.nasa.gov/SST/

This is a blended product (i.e., multi-instrument, and gaps filled) with 1km resolution. There is also a 1km MODIS land surface temperature data product:

http://modis.gsfc.nasa.gov/data/dataprod/dataproducts.php?MO...

Neither of these is real-time (more like daily).

Just for the fun of it, for those of you in the bay area, here's a surface temperature forecast for today, 16:00 PST. It is from the forecast site that I've been maintaining for some time now - http://www.norcalsoaring.org/BLIP/BYRON/index.html

http://raspbucket.s3-website-us-west-2.amazonaws.com/BYRON/F...

(Based on 3km/750m WRF model run two times a day based on the NCEP data. )

I wonder if they are aware of the super high resolution weather satellites that NOAA is putting up into orbit and will come on line, I believe, end of this year. I was talking to a high level NOAA official on the technical side and he was saying that it will essentially provide a remarkable, i.e., revolutionary increase in prediction accuracy and be able to provide on the ground climate level predictions.

Nice work!

There seems to be a typo in the Stats (my emphasis):

>Pixels: 16-bit unsigned ints, representing "deci-kelvin" (i.e., divide by 100 to get the temperature in Kelvin).

It should either be centikelvin or divide by 10.

On another note: It would be cool if you could hover over a certain area to see its temperature.

Please use Celsius or at least show both.

I don't get where the data is coming from. A lot of it must be interpolated.

Amazing stuff! It's things like this that makes me proud to be a developer.

This leaves me wondering, how does one go about designing such a system? Are you planning on doing a technical write up soon?

I didn't realize how hot eastern china was!

In lowest zoom level, if you scroll the map, it snaps back to the original center after scrolling fades out.

It's a cool project, but...

>>It regenerates every hour, providing a constantly updating snapshot of air temperature around the globe.

Once per hour is not "real-time." It's "once per hour."

very cool