Has anyone’s forecasts been really bad lately? The last 3 or 4 days, both temperature and conditions have been so bad, I could swear that they are for another country.
Today for example, it was giving maxes of 11°C, but it only managed to get to 6.6°C.
The GFS and ECM raw data wasn’t anywhere close to 11°C.
It showing temps getting down to 0°C but there’s absolutely no way that’s going to happen. It’s going to be overcast all night and all day tomorrow, but WXSIM says clear skies.
I don’t know what’s going on, but a lucky dip would be more accurate.
Funny that because I was thinking the same. Max temp 6c WXsim 11c overnight -2 actual 1c
Im going to change some of the mix settings and see what happens.
Anyone else?
I have no idea how to get autolearn to work. I tried in the past and gave up.
Kinda getting fed up of WXSIM to be honest. None of the variables have been accurate, particularly wind speeds - they are always miles off. I spoke to Tom about it, but despite sending him files, it didn’t get me anywhere.
Forecasts for my place have been quite off lately, too. Low temps forecast about 4degC lower than actual, daily also not very close to the forecast.
I haven’t changed any settings so far. I’m using 20% GFS and 80% ECMWF and have autolearn running.
Maybe it’s something in the model data that is causing this? Or the global models are unstable because weather is very dynamic these days.
I think we may be expecting too much for WxSim to be correct all the time.
I’ve just looked at the Windy app which allows you to compare the forecast from 7 different models for a given location. This evening the maximum temperature given by any model for my location is 4C and the minimum is 0C. If the big models running on incredibly powerful computers can only agree to within 4C should we expect that we can do any better? Thinking about that another way, if one model was always right why are there so many models out there?
I guess it also depends what you’re trying to use WxSim for. My uses don’t need the temperature (or any other value) to be absolutely accurate. I’m doing one of two things with WxSim output:
Getting an idea of what the weather will be like in the next day or two. Will it be so cold that frost averse plants will need protecting? Which coat do I need to wear when I go to watch the football today?
To try to understand why the weather does what it does. Can I understand particular physical weather patterns as they develop from the way the data looks?
Neither of those need particularly accurate values, just pretty good values.
And if you don’t like the output given by WxSim then don’t use it. Go back to the big models and notice that today GFS was spot on but ECMWF was way off and yesterday it was the other way round.
Max temperatures to be 4 or 5°C too high is not acceptable.
When strong winds are forecast, WXSIM goes completely over the top with 10 minute average speeds.
Another bad habit WXSIM has; When temperatures are below freezing (in my case, the temp was already -2°C) WSXIM runs it’s automated forecast, but says the overnight low is -1°C, even though current data was already below that value, and fell even further than that.
Every other forecast site available is more accurate. So it begs the question, why pay £££ for something that is producing worse forecasts than the free ones?
Are you sure you use the paid version?
The demo version is not localised to your situation.
There are 4 different paid for versions:
Standard and Professional with either
Basic or Enhanced customization.
Until I moved to a city appartement, I lived in a normal village environment.
I had very accurate forecasts and none of the problems you currently have.
I even developed a WXSIM forecast-script to include in my templates.
has anyone got a copy of the wxsim to wu comparison forecast script? that i can have.
Done a search on the forum etc and net no luck.
I want to setup thi s and see what results i get.
thanks in advance…
So what happens when you try autolearn - how have you set it up?
It is a complex program - or appears so. Many things I have found can be left alone at defaults. I use mine with a 3rd party scheduler to launch things when needed.
Well, I didn’t blame the WXSIM but just commented that the temperature forecasts were not really accurate lately. Over the last 2 or 3 years (since I have the WXSIM), the forecasts the WXSIM produces for my place were typically better than the ones from our national weather service.
The “bad” temperature forecast lately were most probably due to often changing model outputs and differences between the GFS and ECMWF. But even more important effect was the wind. Wind forecast for my micro location is nearly impossible as it is conditioned by local topography (and I don’t expect WXSIM to know it). When the cold wind is expected, it does come from the forecast direction but at higher altitude only. When it rolls down the hill, it tunnels in the valley and increases the temperature significantly (up to 5-10°C).
I’ll take some time and try to understand various tweaks the WXSIM offers - in the end, this is a nice hobby and WXSIM is a very powerful tool.
OK, I’m seeing this, and of course want the forecasts to be accurate, as I’ve invested over four decades and many thousands of hours into it. Also, there’s nothing that would happening with WXSIM itself to make it suddenly get worse (it’s been working well here). There’s always the issue of whether or not it obtained fresh data, or whether some setting is inappropriate. Bugs are possible, and I want to resolve any of those that come. Please email me with any specific issues you may have. I’m actually off work tomorrow and will have a good chance to check email. What I almost always want to do first is have you send me these files: localdat.txt, localcal.txt, latest.txt, custinit.txt, wxsmini.txt, wdata.txt, wdatamixed.txt, cty.fdt, rgcode.txt, fdata.txt, and cor.txt (not all of these will exist for everybody), and I try to replicate your forecast.
Thanks for your support, and interest, and for using my program!
Hi Tina, sorry you’re having troubles. I’m happy to fix that, and you can certainly ask me directly. I’ve never seen that particular error (75) with euromix.exe (but it suggests that it’s been set to look for a path or file that’s not on your computer). That will be very simple to solve (send me euromixinit.txt and it’ll probably be apparent what’s going on). That message box is likely indicating that you are NOT using ECMWF data, because euromix.exe is not creating it. I actually am changing how that messaging works in my next (very soon, and in fact the latest beta makes that change), because some users DON’T want to use ECMWF data, and it keeps popping up suggesting it (my new version allows a permanent ‘no’ answer). But yes, you can definitely go with full ECMWF. Please be patient and let me help you fix this. (By the way, I just tried getting GFS and eCMWF data from Chris’ site, and I’m getting an internal server error. This seems to have just happenined, maybe in the last hour. I’ll contact him about that).