Using weather.us to look at all the cloudcover models

[Brad Templeton]

In discussing the weather for GSSP, I pointed to the site I use, weather.us which, while not one of the astronomy-aimed sites, offers some of the most comprehensive information and the ability to isolate cloudcover forecasts and look at many different supercomputer models.  Most sites have picked just one model, but there are tons of them and they differ.  Sadly, one is not more accurate every time.   However, if you find most of the models agreeing you can be more confident.

You can get into the map for GSSP at a link like:

https://weather.us/model-charts/german/1212-w-406-n/total-cloud-coverage/20260713-0600z.html

In this case, I have already used the "Change map selection" menu to isolate the map region I am looking at.     And I've picked the parameter I want to see on the All parameters menu.  In this case, total cloudcover.  (You can also get clouds at different levels, percent sunshine and others of interest.)

Next you have to pick the date and time and the model.   Not all models predict all times, which does present a minor problem.     Once you have picked a time, you see a list of all the different models -- FGS, GEM , ACC, IC and more -- that have a cloudcover prediction for that time.     Almost every model does a 1am prediction if they are doing that date, so that's a good place to start.

Click on the blue boxes to see the prediction for that model.  When you are on a model you can also use the left and right arrows on the date/time box to do a little animation of clouds moving.    As you move to some times, you may see different models become available or not.   Most models do at 1am and 7am many days out.  They also do down to every hour for the next day or two.  10pm is also a popular time (daylight time.)

You will quickly notice models can disagree greatly several days out, and that's because there's lots of error in all the models.

I used this for the last total eclipse.   I found the one spot that every single model said would be clear skies and went there, and got 100% clear skies.    But was I smart?  My first site, in Texas had a scary forecast on all the models with rain and clouds, but they did end up seeing it.  But for them it was luck of the timing, but at my new site there was a lot less luck.     When you are deciding go/no-go, realize that some of the models will definitely be wrong.      

You may want to use the "zoom out" button and see a wider area.   That's because you don't really care about exact cloudcover at 1am in Adin, you want to know how long the clear patches will last, and for that you want to see the bigger patterns of change.  And maybe you like the simplicity of one  core model.  Atmospheric is a good site, it uses the Canadian model (known as GEM on the weather.us screen.)   Clearoutside.com reports it uses an average of several models so that might be easier than using weather.us.  Venutsky also has a nice interface an integrates models.

[John Pierce]

On Thu, Jul 9, 2026 at 1:52 PM Brad Templeton <bra...@gmail.com> wrote:

In discussing the weather for GSSP, I pointed to the site I use, weather.us which, while not one of the astronomy-aimed sites, offers some of the most comprehensive information and the ability to isolate cloudcover forecasts and look at many different supercomputer models.  ...

 

Atmospheric is a good site, it uses the Canadian model (known as GEM on the weather.us screen.)  

Astrospheric Pro shows 3 models plus a blend,  RDPS (the Canadian model), ICON (German), GFS (the US Global Forecast System), and NBM (a national blend)

See https://www.astrospheric.com/DynamicContent/ensemble for an explanation.

The Pro version is $3/month, or it can be made available to an 'astro society' for free if that society features astrospheric on their website's home page.

[Mark]

John,

TAC has a website.  And it is an AL member club.  Perhaps it should feature TAC-AL and Astropheric on its web site home page and make the pro version of the weather app available to anyone subscribed, if possible.