New WXSIM Version coming soon!

Hi Everybody,

I’m working on a (fairly modest) upgrade/update of WXSIM. The main impetus for this has been to accommodate the new Cumulus version which I think has started beta testing. It has a new log file name structure, which affects WXSIMATE, wret.exe, and WXSIM-Lite. In each case, I’m simply adding the option of using the new version of Cumulus, without taking away the old one. I’ve also tried to accommodate use of Unix-format files, for people running WXSIM on Linux, but this works only partially at the moment and the better course is to simply have some little utility convert the files first. If you are beta testing the new Cumulus and want to try WXSIM with it, here’s the installer for the WXSIM (etc.) update:

www.wxsim.com/wxsimupg2024beta.exe

To change to the new Cumulus, in WXSIMATE, go to Customize/Local Station Source. In WXSIM-Lite, go to Station Software. In wret.exe, it’s a little complicated: you have to select a forecast to view, click the Compare to Actuals button, say either Yes or No (either is OK) when the message box appears, and then make the change at the bottom of the incomplete graph that should show up, and Close that and try again. In each case, the new Cumulus is the one with the numeric log file names.

Note that the associated readme files have not been updated yet, so you can ignore them.

Another change I’ve made (a good one, I think, at the suggestion of a new user who understandably got confused) - but is not reflected in the available download - is that I’ve changed the ‘Run Immediately’ check box on the Auto Run and Other Setting form to a simpler ‘Run Immediately’ command button. That’s probably more straightforward.

If there’s something in the operation, documentation, or presentation of any of these programs, which you think could be cleared up in some simple way, let me know. I know the system is en “eyefull” and can be daunting to a new user. Some suggestions might be easier to implement than others, but I’d like to know if you have something in mind.

Also, ideas for new features or outputs are welcome.

Thanks!

Tom

Great… Thanks for the update on the new version for Cumulus MX.
Any Ideas for output etc I will let you know

Hi Tom, great to hear about the update. Regarding new features, I have one. If I recall correctly the WXSIM-Lite can calculate corrections based on real data vs forecasts coming from the GFS “part”. If one is using e.g., 80% for ECMWF, these corrections are not as useful as they could be, right? Can this feature be extended or is it too complicated?
Thanks!

I have yet to find a good mix of the GFS V ECMWF to get the forecast near what the station is saying, most time its a little off, but I dont mind

You’re right. At present WXSIM-Lite uses only GFS data. It’s possible I could extend that to ECMWF (or a mix of them), but it would be a fairly big task, which I’d probably need to work with Sam (Bohler) on. It would likely produce a quite small, but perhaps measurable improvement. Meanwhile, the last time (2-3 months ago) that I tallied up the effects for different mixes of WXSIM-Lite with WXSIM, the sweet spot seemed to be at about 50% usage, as opposed to the 65% or so I was recommending earlier. It’s still providing benefit over WXSIM (which was using 70:30 ECMWF:GFS in those trials), but lacking ECMWF drives it towards a slightly lower optimal usage. I’ll check on that again soon, as I’ve got a few more months of test data (running three different weighting scenarios in parallel).

Hi Tom,
I have installed the Beta versions and also gone live with CumulusMX V4.
Tonight was my first WxSim-Lite analysis and it seems there is an issue with the way it searches for month logs. format yyyymmlog.txt

All runs fine till it gets to 202312log.txt and then results in error 53 (file not found) stating it can’t find 202412log.txt

By chance I was awake at 02:00 and clicking OK to the error the analysis completed, however it does it every time (the second pass) and I re-ran it and same issue.

Rather than roll back now I will need to be up at early hours each morning to click OK :slight_smile:

UPDATE : Hi Tom, Stuart sent me an email and correctly suggested I change my WxSim-Lite Span Days from my normal 90 Days (I reduced it to 80 so it does not look back to last year for now until a fix is found).

Also off topic, Bohler GFS Run 2024-03-26-06z (Lite/Data/Adv) all had issues, McMahon was OK.

First +0 hour was incorrect.

Bohler:
2024-03-26-06
+0,0.0,0.00,50.0,144711.3,269.8,19.6,0.0,-273.2,-273.1,78050.9,-271.9,-293.2,-293.2,-273.1,-0.0,0.0,183.4,1.1,24135.0,-20.0,0.0,0.0,0.0
+3,1017.3,0.00,2.5,2.4,35.7,145.8,5723.5,13.8,11.9,10.0,6.2,2.2,1.2,1.5,54.7,58.8,62.8,70.5,79.3,48.8,25.8,26.4,60.4

McMahon:
2024-03-26-06
+0,1014.1,0.00,0.0,1.8,77.5,119.5,5693.4,15.5,13.6,11.8,7.9,3.7,-0.4,-0.7,47.0,49.7,52.4,60.1,71.5,82.7,41.8,19.3,47.8
+3,1017.3,0.00,2.5,2.4,35.0,145.8,5723.5,13.8,11.9,10.0,6.2,2.2,1.2,1.5,54.7,58.8,62.8,70.5,79.3,48.8,25.8,26.4,60.4

msglog.txt
Beaumaris
26/03/2024
11:10:02 PM
Implausible sea level pressure value (0) in GFS/ECMWF data. Replaced with fairly accurate estimated value (1002.4) using 1000 hPa height and other data.
*
26/03/2024
11:14:06 PM

Going back to bed now :slight_smile:

Kindest Regards,
Tony

Thanks, Tony! Just saw this (Stuart pointed it out … I’d missed it earlier). I’ll look into these things today.

Tom

I just tried a scan spanning the end of the year, and it went smoothly (this is with Cumulus, and the newer format files), with no errors. Clearly, you saw one. I don’t yet know how that happened. Can you replicate it, either by doing a longer scan period, OR by setting the date on the right manually back to something less than 80 days after the end of the year?

If you still see the error, I may need to ask for some files to try to replicate it.

As for the Bohler data, I see you did have an error (on March 26), but I got just up to date 12Z runs from today (4/28) and all seem fine. Not sure what happened on that day!

Thanks again for all these reports (even though I’m sometimes slow to respond!).

Tom