I am currently experiencing a few problems with my forecasts. The temp jumps up about a 1.5C and then lowers again and the pressure jumps and then settles back down within two hours. I have attached lastret.txt that shows the problem. It is also very obvious when you look at the graphs generated.
It has been doing this for some time but never really paid attention to it until I realized it was affecting my WSIM/EC comparison script accuracy.
The pressure on the Davis Console is set for the same as the local weather station (CYXE) and I am not currently using any bias corrections.
That’s a big pressure jump! I’d like to find the reason and solve the problem. To do so, I would need the custinit.txt, localdat.txt, localcal.txt, latest.txt, and wdata.txt files from right after such a forecast was made.
I seem to have nailed down where, but not why, my forecasts have been having problems with the few first hours as reported earlier. I was doing a manual forecast and on the Interrupt Planner I clicked ‘Use Model data from Ready or GFS’ and noticed that the graph never changed. I then clicked on the ‘Clear All’ button and then the ‘Use Model data from Ready or GFS’ and noticed a difference at the beginning of the graphs compared to earlier. Interestingly, there was no difference after about two hours on the graphs.
I have attached pictures of the Retrieved Data text before, the Interrupt Planner before Clearing, The Interrupt Planner after Clearing and clicking Use Model…, and the Retrieved Data text after.
I have tried deselecting the Save for next run box with no effect. I have tried it with and without importing local data with no difference in results.
All this was done from a fresh install of WXSIM on a new computer. If you want, I can zip up the entire WXSIm directory and place it on an ftp server for you to download, unzip and make a few runs on your machine to see the same effect.
One other thing that I just noticed. When I do a Clear All but before clicking Use Model data, there is data remaining on the graph to the extreme left. See the picture attached. I am not/will not be using DST (Saskatchewan never changes it’s time) and all these boxes are unchecked on WXSimate and WXsim.
I realize that this data is ‘Start Time data’ as mentioned in the WXSIM help document but what start time data is this from? The imported local/downloaded data? Sometimes the data right after the start time data is quite different from the next data, which gives quite a jump as demonstrated in the Interrupt Planner before Clear.png in the posting above.
Thanks for that useful feedback! I’m thinking this problem has something to do with repeated runs, or at least with the program remaining open. I have a feeling that if you close the program and restart it for each forecast, it won’t happen (might be worth checking). Anyway, it’s a hint I can follow up on, and I’ll let you know if I need more files.
I have a feeling you are right because after closing WXsim, updating the exe and then restarting, all works fine right at the beginning. The only problem I have with doing this all the time is that when auto starting WXsim I have to put all my precipitin/snow settings back in, turn on the initialize with previously forecast snow cover and then enable the auto run. I’ll verify this tomorrow when I have more time.
If the Interrupt Planner (in auto mode) would Clear all the settings by default and then load them again as I can do manually then this may also solve my problem.
I had a chance to do as you suggested and look at the wxsim tables/graphs and again after restarting wxsim and a run had completed. What I have seen is that indeed a restart of wxsim will clear all data on the Interrupt planner. This can be seen by the results I got in the attached pictures, especially the SLP (the pictures also show the cumulus data for reference). My SLP is the same as the local weather station where Metars come from.
One quick question - On the Interrupt Planner, what is the purpose of the ‘Save for next run’ check box? My thought process is that if unchecked, this would be akin to ‘Clear All’ and then ‘Use the Model Data from Ready or GFS’.
Thank you so much for your investigation of this problem! That should really help me figure out and correct it.
You may be on to something with that “Save for next run” check box. Its purpose is to let you do repeat runs without re-importing data, and I do this often. However, it is possible that something about it could be causing the error. Could you see if UNchecking the box fixes the problem? (My apologies if you’ve done that and I somehow missed it in reading your posts - it’s late!).
I did a manual run, cleared the data (which ticked the ‘Save for next run’ check box automatically), imported the Ready/GFS data and ensured the ‘Save for next run’ check box was unticked. I then completed the run manually.
An hour or so later, I watched an automatic run and saw that the ‘Save for next run’ check box was indeed checked again. I am running WXSIM on an older HP mini (atom based processor) so it does not run the fastest but slow enough for me to see all that is happening.
I’ll monitor it for a while to see if my pressures/temps etc start getting out of wack again.
I hope this helps.
Carl
Edit - After 4 automatic runs, I can confirm that there was no change to my symptoms after unchecking the ‘Save for next run’ box.
Playing with the program here, I’m so far unsuccessful at getting the pressure spike to happen, but I know you and at least one other user I’ve heard from have been seeing it.
If you can get it to happen on a manual run, and then immediately send me the files I listed above, I might be able to replicate it. I’ll also need to know exactly what you did. Actually, a ‘Run Immediately’ auto run might sreve the same purpose. I just need to be able to replicate your sequence of actions exactly.
Even better than one run, tomorrow I’ll restart wxsim and do 4-6 automatic runs one hour apart (saving all the files for each run) to simulate my hourly forecasts. Then I will zip them all up and get them to you. This should allow you to see the pressures/temps progressively getting worse each hour.
I can still zip up my entire c:\wxsim directory (minus the many existing forecasts to save space) and get that to you. Then you could rename your existing c:\wxsim directory to something else and then use mine to try it out to see what is happening.
I will definitely do the 1. tomorrow and can do 2. fairly quickly if you like.
Your first idea would be great, and even just 2-3 runs showing the problem should be sufficient.
I’m curious if you can trigger the problem with a set of successive runs on the SAME ingested data, or if it has to be different data. I’m trying to figure out if it’s some artifact of left over data from a previous run, or of actually being run again on the same data. For instance, if you do a manual run (or ‘Run Immediately’ auto run) and then do one or two more of them with the same data, does the problem appear?
I did as you suggested above, I ran forecasts with the same set of data 6 times and the results all stayed the same, i.e.no difference in results.
I also sent you an email with 5 sets of sequential data that I can make show the unwanted result. It was emailed to you because of the file size/number limitation.
