Cumulus Support

I noticed my wxsim forecast data didn't change during it's scheduled run. I ran a manual one, and it's giving me an error.

Chris has server and database issue which he is working on so the site may go down and if up may certainly not be stable until he has things sorted.

Stuart

Ahh

It's been cooked for about 10 hours for me.

Check the latest forecast out.. That'd be nice!
I disabled all ECM stuff, and set it to Bohler, but I am still getting weird forecasts like that

Still no data from Chris' server today. My forecasts have just stopped. Using Bohler data should work as long as you completely turn off ECMWF data and merging and reset WXSim to only import GFS data. I'll try that on mine this morning and see what happens.

Stuart

Just run a forecast using Bohler data and no ECMWF and it looks fine so for me Bohler is working.

Stuart

It's the "Use bias-corrected model data" setting that is causing the silly forecasts.

Despite WXSIM-Lite having 14 days of data to mess with, it's not working right for me.

Do you see the bad temps on the wxsimlite graphs when you run it manually?

Stuart

This is what I get after a full scan
For whatever reason, it always stops at 04/02/2025 18Z

And then when I do a forecast, I get this:

That fdata.txt file is updated on every run of WXSimLite from the normal schedule prior to a forecast run, so something is amiss there!

As for the scan if I remember correctly that is what the graph looks like after a run, if you watch it from the start it should show a starting date of -60days in your case and work forwards from that. I just started a full scan manually and yes mine started at 27/11/2024 as mine is set to 70 days.

So if yours actually starts at 04/02/2025 then it is not finding the data it needs.

My 15:00hrs auto forecast ran fine using Bohler data and Lite fdata.txt and shows reasonable forecast.

Stuart

It has just ran it's scheduled daily analysis, and again, it shows 04/02/2025 18Z

I don't understand why?! I have a 14 days of all 4 runs from the 26th January, up to today's 12Z, but each time the analysis is ran, it always ends up at 04/02/2025 and that makes the fdata out of date.

Have been getting WXIMATE erros with McMahon.
Thx to this post in the forum, I understand now the root-cause.
So switched off ECMWF and selected Bohler. After that no errors in WXIMATE and automatic forecast now running with GFS/Bohler.

https://leonas.ddns.net/Wetterstation.S ... recast.txt


Waiting for weather-watch.com site to come back before switching back to ECMWF/McMahon again.

lse-cumulus wrote: ↑Sat 08 Feb 2025 5:13 pm Have been getting WXIMATE erros with McMahon.
Thx to this post in the forum, I understand now the root-cause.
So switched off ECMWF and selected Bohler. After that no errors in WXIMATE and automatic forecast now running with GFS/Bohler.

Waiting for weather-watch.com site to come back before switching back to ECMWF/McMahon again.

Yes that's what I've done.

I sometimes forget to turn of autorun in WXSim and change Import file name to wdata.txt from wdatamixed.txt or whatever names you have used. but I soon remember

Stuart

Mapantz wrote: ↑Sat 08 Feb 2025 5:03 pm It has just ran it's scheduled daily analysis, and again, it shows 04/02/2025 18Z

I don't understand why?! I have a 14 days of all 4 runs from the 26th January, up to today's 12Z, but each time the analysis is ran, it always ends up at 04/02/2025 and that makes the fdata out of date.

OK so to refresh my memory I just ran a Lite full scan. At the end the last opened file was the same as yours 4th Feb but as I pointed out it started at 27/11/2024 which is minus 70 days (for me) so what you see at the end IS correct, what you need to ensure is it starts at at minus 60 days. Also fdata.txt is NOT created by a full scan it creates cormod0.txt. the fdata.txt is created by a run immediately prior to a forecast run by the normal schedule. Also make sure you ahve the correct data in the CMX logs and wxsimlitedata directory.

Stuart

I don't have any data prior to the 26th of January though.

The problem with fdata.txt not being up to date is that the forecast won't run unless I click OK. That's no good for the automatic forecasts.

I also ran a forecast with all the wxsimlite stuff disabled and then did one with it enabled (clicked OK on the box about fdata being out of date) and compared the two - they were both identical. Hence why the popup box says it won't include any data from fdata.txt.

Also, what is a full lite scan?

OK, you have just enough data to run a scheduled scan in WXSimlite (minimum is 14 days of forecasts in the WXSimLitedata folder) which you can schedule to run one per day and that will create a file called comod0.txt. Also on that schedule screen you set hours to run forecast in auto mode so if you run forecasts at say 06, 13 and 18 hours then schedule that with minutes set to make sure this completes before WXSim itself starts, this will create fdata.txt and will be used by WXSim. If you only run manual forecasts the obviously run a download & go manually onl WXSimLite prior to running WXSim.

Either of these methods should give you a valid forecast. Note that McMahon data is not used by WXSimLite this ONLY access Bohler data.

Finally this works fine for me!

If you still have issues I suggest you contact Tom Ehrensperger for help.

Stuart