Bug with Wind Gust??? **solved**

[lambjon]

Hi,

Completely new to the forum, and installed Meteobridge for the first time yesterday. Been pushing data to WUnderground, but noticed some strange readings for Wind Gust data. I appear to be getting a gust speed of 11.3Kmh frequently when the wind is totally calm outside. I wasn't sure if this was a bug on WUnderground initally, but I've been able to track it back to the Meteobridge sending out this erroneous data by executing http://meteobridge/cgi-bin/template.cgi?template=[wind0wind-max2=kmh]%0D[wind0wind-max10=kmh] in the browser. When I do this, I'm seeing the figure of 11.3 shown at the same time it appears on WUnderground.

I have a Vantage Vue station and can confirm that neither the console, nor Cumulus software is picking up these gusts (because they aren't actually happening!!!). Any ideas???

If you view the history data here (http://www.wunderground.com/personal-we ... T6#history), you will see the problem if you look at March 14, and view data around 6:30pm-8:30pm. During that time, the wind was totally calm, and a max gust of 1.4kmh was recorded in Cumulus for that entire duration.

Any help would be appreciated. I've even tried writing my own HTTP GET statement, and I got the same results as using the built in Weather Networks option.

Kind Regards
Jon

[admin]

We never had that finding before. No idea why Meteobridge should invent gusts not being there.

Are you using Meteobridge in "pass-through" mode, where a PC program makes use of the
Meteobridge as if it is a Davis IP logger?

[lambjon]

Certainly was strange. Yes, I was using "pass-through" mode to observe live readings in Cumulus when this problem was occuring. Can not say whether or not opening that software on my PC caused this problem, I'll have to keep an eye on it. What I did do was reboot the Meteobridge and that seemed to clear the issue. If the problem returns I'll try and work out what may be initiating it.

I'm still getting used to all of this at the moment. For example, I noticed that when using the "Weather Networks" feature, updates to WUnderground were not always consistent in their timings (ie: not always 5 seconds), sometimes a full 30 seconds may pass before an update. Is that by design because the data may not have changed? If I create my own HTTP GET line instead, it is much more consistent.

Either way, I'm very happy with what Meteobridge is able to do

[admin]

For example, I noticed that when using the "Weather Networks" feature, updates to WUnderground were not always consistent in their timings (ie: not always 5 seconds), sometimes a full 30 seconds may pass before an update. Is that by design because the data may not have changed?

Yes, exactly. It spares bandwidth, when nothing new to upload. But uploads are done once a minutes latest, even when nothing has changed.

[lambjon]

Just to follow up on this, I've experienced the same problem again, and this time I can definitely confirm that the problem was caused from the point I started using pass-through to view data in Cumulus. I observed erroneous wind speeds approximately once every ten minutes. What I saw happening was an incorrect instantaneous high wind speed (in todays case it was 14.4Kph, so different from the other day), which was then of course being held as a recent high gust value. This continued until the point in which I closed the Cumulus software (so the Meteobridge reboot I did the other day was not in fact required).

The issue is not immediately reproduceable though, as I've had Cumulus running on several occasions since I first reported the problem, and not observed any erroneous wind readings being sent from Meteobridge. Indeed, after I noticed it tonight, I've since reopened Cumulus, and data from Meteobridge appears to be transmitting accurately.

I have not observed any erroneous data being transmitted from any other sensors, it only appears to be affecting wind data. It's a very strange problem.

[lambjon]

Actually, on closer inspection it is doing it again right now with fake data, only this time the "fake" speed is 6.4kph. So there is a chance it is always happening, but sometimes the fake value is a low one and therefore not affecting the information stored in history on Wunderground and other weather services (because there are other "real" higher wind gusts).

[admin]

In order not to loose any max gust data by not having polled exactly the second where the gust appeared, Meteobridge also
inspects LOOP2 packets and incorporates "10 minutes max gust" data from there.

When Cumulus is not working with wind max data from LOOP packets, then it might not grab these hidden max gust values,
which explains differences between what cumulus reads out from the station compared to Meteobridge.

[lambjon]

I understand what you are saying about the gust data, thing is, what I'm seeing in Cumulus matches exactly what I see on the console.

I'm also however seeing the erroneous data on the [wind0wind-act] template (if I happen to look at the right moment). I don't understand why it is outputting a "live" speed of say 14kmh, when both Cumulus and my console has no wind readings even close to that. When this occurs, if I query [wind0wind-max10], it will also show a speed of 14kmh (to be expected), however, if I look on my vantage console and look at the reading for Max gust in last 10 mins, the value will not match in these conditions (eg: earlier, Meteobridge said 14kmh, the console said 3.2kmh)

[admin]

I am out of ideas why this happens in vantage pass-through mode.

[lambjon]

That's fair enough. I do some computer programming myself (clearly nothing to your level), but I cannot see any pattern in this that could help to track the problem down.

I will just ask though, is it possible to output all recent readings from Meteobridge in some way?? For example, when I get this erroneous wind data occur, I can see the same fake value pass through the selectors that look back in the past (eg: [wind0wind-val2]). I'm assuming therefore that Meteobridge holds a data set containing every sensor value for the past hour?? If there is some way of outputting all that information in a long list, it may be possible to spot a pattern as to when this fake reading is occuring?

If there is nothing that can be done, at least I know what is initiating the problem, so I can keep an eye on when it happens, and delete erroneous data from weather networks as necessary.

Will be interesting to see if anyone else experiences this at some point.

[admin]

Can you please tell me the MAC of your Meteobridge. I would like to include it into my beta tests for
next release and by that I can apply some experimental code changes which might fix or
at least make more obvious what the root cause is. Please send MAC to info(at)meteobridge.com.

[lambjon]

I've sent you my MAC address as requested.

I don't regularly reboot the meteobridge, so let me know if you need me to do so at any point to pick up a beta version.

Today has been a very calm day outside in terms of the wind, which makes it easier to spot the fake readings. I noticed earlier that I seem to be getting 1 fake reading every 10 minutes, it is in fact almost exactly 10 minutes apart too, but was as much as 10.5 minutes sometimes???!!!!

I'll continue to provide as much info as I can if it is going to be of any help. I can for example tell you that I've observed the following:-

Whenever I am getting these fake readings by having Cumulus running, the fake value being sent is always consistent for that session. (ie: if it gives me a fake speed of say 14.4kmh, it will always give me that same value until I close and restart Cumulus). Upon restart, if the problem happens again (and it does not seem to occur all the time), I will get a different fake value being generated (so it could be say 3.2kmh), but it will remain with that value until I once again close Cumulus.

[admin]

yes, please do a reboot, so that it loads the new code and give that a try.

[lambjon]

Rebooted earlier (now have build 6130) and so far so good!

Did you find something specific then that you thought might be the root of the problem?

[admin]

Looks like asking for "max wind gust" in LOOP2 packets does not work the way cumulus is polling data.
I still have no full understanding but at least a clue. I changed only this.