Wind data not consistent (UDP vs Web)

Today I’ve noticed the wind data (avg, gust) that weewx reads from the UDP is different to what is shown in the blue web interface. I haven’t noticed this before yet. I am quite sure the UDP data is the “real” one since I have a DVP2 just next to the Sky (see also comparison plot for the last few hours). Is there some processing done on the server side perhaps?
The “blue” page:
The weewx page (UDP data):

WF Sky vs Davis comparison to prove the UDP is correct:

Hi @petercek,
I can not see the difference that you mention. The average and gust value seem to match. It would be clearer if you specify a single 5 minute period to compare.
If there is a difference in the 5 min values then burrow into the one minute values that contribute to the 5 min period to find the discrepency.
In the past I did find an error in the one minute values were not calculating all the 3 sec values into the one minute total correctly. So it is possible that mistakes creep in.
Cheers Ian :slight_smile:

The Sky is in the power save mode (15 sec period). The difference is in the gust speed, 10km/h at least. Weewx’s max gust was 59 km/h, the blue page reports 69 km/h. Regardless the calculation window, values don’t match.

There are calculations done on the server side. In the past there was an error how Weatherflow calculated the one minute value from the 3 second values. The method I used to check the calculations and find the error was by videoing the 3 second values live and working out the correct one minute value. Once the error was discovered they fixed it. I have not checked it again since then. And if your readings are different to 3 seconds then the Weatherflow calculation will be different.
My weewx and the Weatherflow web graph are currently showing exactly the same values for wind gusts.
Cheers Ian :slight_smile:

Thanks, Ian. I will then rely on the UDP data.

I think they are both “correct”. At least the average wind speed is. However the graphing system does a weird job when it comes to the gust value… when you zoom out, a single data point on the graph (and it has only a measly 32 of them visible at any point) is calculated from many data points. In your graph every point uses half an hour of data. The average wind speed is calculated by averaging the points during that period, but the gust just takes the max value in that half hour period. I have mentioned it before, but this isn’t the best way to handle the gust data. For lull it does the similar thing, but taking the minimum, which almost always ends up at zero when you zoom out a lot.
Gust should be defined as the max during a fixed period (say the last 10 minutes, there are multiple definitions going around). But when zooming out it should average those values, not take the max again.

If you fully zoom in, you will find that the udp data matches the graph.

I did zoom in and found that problem. Even if the problem was only in graphing, the History data tab should contain the right value, as I read it from the UDP but it does not.

I think you should check that weewx is displaying the correct values. There is a mistake somewhere and it could be anywhere until you actually check what the value should be from your 15 second values.
You only need to do it for 5 minutes or a little more on each side of the 5 minutes to be sure.
I would just sit there writing down each 15 second wind reading and calculate what it should be. That will confirm which values are correct. And if you video the readings for the 5 minutes you can play it back quicker and pause as required. That is how I did it.

As I said, I have a Davis Vantage Pro2 mounted 30 cm from the Sky and running 2 instances of weewx. The weewx values match more or less. This is how I confirm the (wf) weewx reads the correct values. The WF web values don’t. This is what I noticed just recently.
Thanks for your suggestions, I will just leave as is. I just wanted to mention here if anyone else has noticed the same or if there was an explanation for it.

could you send us a snapshot of the fully zoomed in data for the weatherflow and for the weewx?

I cannot zoom-in anymore, it’s too late now. But there is fresh data from this night. Not so significant but it does not match.