If you live in North America, or (to some extent) UK and Australia you have local atmospheric seeing forecasts available, like the Canadian
If not, or you are going to that exotic place of yours (astro-farm/resort, or just grabbing your airline-friendly equipment), you should probably look for global options.
When I started using astronomical seeing forecasts a few years ago, this was mostly for my locations in Norway, and a few preferred destinations like Cape Verde. Anyway, I had to rely on global forecasts.
I checked out the free 7Timer (http://7timer.y234.cn) back then, but didn't like the interface much and it looked to me that the Swiss commercial service meteoblue.com also provided better precision for my locations.
I didn't test the 7Timer for long though, and when I hit the link a few months back the site did not respond (not even from the last known IP-address of 18.104.22.168). It's been like that since. The Astro Panel app (last updated 2014), relying on 7Timer data, has been informing for a while that the provider (7Timer) should be up again "in a month".
You do have the freely available GFS-data (NOAA/NCEP weather model) that you can chew, and some sites have presented various visualizations (like http://wxmaps.org/pix/euro.jet.html) from such.
But this review is about astronomical forecast services and assume some sort of user convenience like easy lookup with site precision and checking at a glance.
With 7Timer down (for good?), that leaves us with meteoblue.com alone for the remaining review. (any other service out there would be a well-hidden secret)
I have used the meteoblue's free and ad-based option for years, providing me with the same forecasts as their paid option but limited to 3 days. I did once have their 14-day trial of their premium service (longer term forecast, no ads), and that would potentially help me in choosing weekends for driving far with lots of equipment.
So last autumn I contacted meteoblue about a (budget) subscription for astronomy only. Their point+ service opens for all their special forecasts (agro, snow, astronomy etc), but is pricey at €50/yr.
To illustrate my hesitation, the wind forecast service WindGuru Pro costs me €17/yr, This one has comparably more features over the free version, and it offers higher total value considering price.
I did express my opinions to meteoblue. They elaborated their dilemmas, and provided me with an extended 6 month trial so I could the service quality over some time. I have now just ended this period, and will share my findings.
This winter provided only a few good nights, and I could of course not get out every time. I'm mostly relying on mobile imaging, and the transport requires a bit.
So it did not make much sense trying to present a rather small data-set. Still, I hope my opinions and impressions will help.
Their web site is not using all the latest technologies and designs. It's not responsive, and does not fit well even for a medium sized tablet with hi-res screen. But quite nice looking and usable, way better than the 7Timer site.
meteoblue's not so tidy apps (tried Android) are merely useless though, at least for astronomers, since they never incorporated the special forecasts.
Most hobby astronomers that I know nearby do not care too much about looking up up seeing forecasts. The cloud cover is what's relevant for them. Hoping for clear skies when having some spare time. Up here autumns can be quite good in terms of transparency, late in spring (May) is good but too bright for DSOs, and the winter months in between has crazy much clouds. That is, most of the time our struggle is clouds. I choose to mention the cloud cover, as meteoblue guys do forecast this lined up first.
The public no-ad weather forecast service yr.no financed by the Norwegian government. It seems to be generally be the preferred weather service all over Scandinavia. Not surprisingly, as their staff had around 80 meteorologists in hire last year. So, they are able to do a lot of tweaking for their models. (you can actually see that the precision drops a little Friday afternoon when they go for a beer or whatever)
Yr also has detailed cloud forecast, and provides a free API if you want to extend the calculations or presentation.
meteoblue did not come close to Yr regarding prediction of cloud cover. That does not mean it is worthless at this point. I prefer checking several services to see "opinions" (uncertainty) as none of these individual services present alternative outcomes anymore. (dmi.dk had their triple-scenario presentation some years ago, that I found quite helpful for planning)
Roughly, when disagreeing with Yr, meteoblue turned out "right" one of five times.
The seeing forecast is meteoblue turf though.
It is only recently I started scanning the "Bad layers" columns, so and have not made much use of these values yet (can't guess about their precision). Likewise with "Jet stream", although high speeds above 40 m/s surely seems to imply bad seeing. On the other hand, really low speeds (< 5) did surprisingly not seem to exacerbate seeing, both as forecasted and as seen.
What I did follow closely whenever I had a chance, was the three seeing columns: seeing1, seeing2 and arcsec.
High values if seeing1 (rare!) seemed to be a hit, turning out great nights especially if seeing2 was high too. I could not make use of the seeing2 independently though (weight on flickering, according to help), seeing1 seemed to rule all the time.
But the arcsec column captured my interest, being some calculation of seeing1, seeing2 and arcsec.
I'd like to think it actually "worked" an helped me out. But as said, I have not been able to jot down enough "results" over the short and rather cloudy period. Thinking about it, now with fast USB3 planetary cams, and sensitive/short "lucky" DSO imaging, this measure could be quite convenient for imagers.
Nice. I noted on a couple of occasions when going to a high site, almost below bad layers for once, this showed up different (better arcsec) compared to a nearby lower site, as confirmed later. But again, not enough data, might be a coincidence.
Not sure. What I did not like is the secrecy about the calculation of seeing1, seeing2 and arcsec. It is of course meteoblue's right to not choose disclosure, and I would guess it's for business/protective reasons.
But since it looks like these measures/scales/approximations are inventions out of meteoblue heads alone, and does not seem to be widely used at all, I'd say we'd like more than the short help page.
For instance, providing arcsec value is almost too good to be true. And is definitely not always hitting, so please let's have some assuring guidance. I guess some will choose to have GFS jetstream data directly, more so if the values are obscured.
Can't yet say how much the plus-3-day seeing forecast is worth. As expected, the forecasted days usually changed a lot in a couple of days, just like the various cloud cover forecasts. But definitely some planning help.
From here. For me, the 3-day free meteoblue may be good enough, event though I do mobile imaging or travel to something more permanent (not automated) at my cabin in the mountains. Right now, €50 is too much for not getting much more than the free product. But my needs may of course grow.