Problem with windspeed graph

[DaveNZ]

Hi, I have just used cumulus to set up a site for my weather data here: http://howickweather.orgfree.com/weather/trends.htm

I turn my computer off at night while I sleep since it's rather noisy, and cumulus just downloads all the data from the station datalogger the next morning. When I looked at my windspeed graph it isn't right. Between about 1am-2.30am and 4am-6am the average should be on 0 (it doesn't seem to have a problem going to 0 when computer is on). It's not just the site, also the graphs produced on my computer.

Any suggestions as why this has happened?

[steve]

Was this just last night, or does the wind drop to zero every night where you are? Could you let me have an example of this? I guess you will have to wait until the next time this happens, as you have your graphs set to 12 hours, so they're only displaying the daytime figures at the moment.

Could you also let me have your Dec09log.txt file at the same time as the graph, and tell me whether you have 'calculate 10-minute wind average' set, and what your station's logger interval is set to?

You could select the relevant period in Select-A-Graph; that basically does the same thing as at start up - it draws the graph from the data in the log file.

Note that when data is taken from the VP2 logger, the only "average" available is the 3-second "instantaneous" wind speed, so that's what Cumulus has to use. See the other (long) thread regarding this.

[DaveNZ]

Just last night so far as I only set it up yesterday. I took a screenshot of it which I will attach. And yes I did read the long thread you are referring to and so I know all about the wind speed discrepancies when Cumulus is closed and then restarted if "10 min average" is selected

But the problem was that early in the morning the real wind dipped to calm - but the graph is plotting it at about 1.6km/h. Cumulus was restarted about 8.45am this morning. At the moment to avoid the problem mentioned earlier, I set the calculate 10min average off before I close cumulus and back on when I open it again (but only just started doing this this evening). Archive interval for logger is 1 minute.

[steve]

It's very odd; it does look as though it's supposed to be getting down to zero, but it levels out at 1.6. Perhaps there wasn't actually any data between those two points, so Cumulus just repeated the 1.6.

Could you run Weatherlink and see what values it reports between those periods?

[DaveNZ]

In weatherlink it's showing up as calm (in the records and the graphs):

[steve]

I'm afraid I can't explain it then. Even ignoring the 'average' issue, the value that Cumulus stores for 'gust' is the same as that 'high wind speed' figure that Weatherlink is plotting. One difference between the way the two work is that Cumulus uses the Vantagepro.dll, which does the conversion to km/h automatically. But all Cumulus does is read that value via the DLL and log it. So it either really did get that value, or it got an error value and continued to use the previous good value. Unfortunately I have no tracing of this as I have with other types of station, as it's not usually a issue.

I would recommend running with Cumulus not calculating the average at all times, by the way. The station calculates the 10-minute average when it's running anyway, so it's pointless switching Cumulus over to calculate it. You may also find that you get odd values at the point you switch between the two methods.

[DaveNZ]

Ok thanks. Well it's 11.15pm 9 Dec here and currently completely calm - will switch computer off soon and see if I get similar results again tomorrow morning.

1 other thing I just noticed is that for some reason cumulus has recently changed the lowest value on the scale of my temp/dewpoint graph to -2.0ºC (it was at about 11º before). Not sure why it has done this as it's not necessary and it visually flattens out the trends.

[steve]

1 other thing I just noticed is that for some reason cumulus has recently changed the lowest value on the scale of my temp/dewpoint graph to -2.0ºC (it was at about 11º before). Not sure why it has done this as it's not necessary and it visually flattens out the trends.

The graphs auto-scale, so I don't know why they've done that either. They do slightly odd things from time to time.

[steve]

In weatherlink it's showing up as calm (in the records and the graphs):

Just to be clear about this: ignoring the 'wind speed', the 'hi Wind Speed' figures are showing up in the WL log as zero, and not as 'blank' or error codes (I can't remember what WL displays for errors values)?

[DaveNZ]

Hi Steve,
well this morning it has done the same thing again, it doesn't seem to be accepting accept the 0 values from the datalogger. In the weatherlink records, it appears as below.

weatherlinklog.PNG

Also I found out why the temperature scale has dipped to -2 - cumulus seems to be doing something strange to some of the windchill values (attached graph). All windchill values are correct in Weatherlink.

windchill.PNG

[steve]

well this morning it has done the same thing again, it doesn't seem to be accepting accept the 0 values from the datalogger. In the weatherlink records, it appears as below.

Can you show me the corresponding data from Cumulus? I'm putting some tracing in, so we can see exactly what Cumulus is reading. The only thing I can think of that is happening is that Cumulus is getting an 'error' figure from the DLL rather than zero, so it continues to use the last good figure. Looking at the charts we might get a rare period of calm here later today, so I can try it to reproduce this myself.

Also I found out why the temperature scale has dipped to -2 - cumulus seems to be doing something strange to some of the windchill values (attached graph). All windchill values are correct in Weatherlink.

Is this 'live' or from the logger? Do you have Cumulus set to calculate wind chill?

[DaveNZ]

I just reset the graphs to 24 hours on my page here so you can see it: http://howickweather.orgfree.com/weather/trends.htm
Here's the data for the period concerned last night (just copied from Dec09 log).

cumulus10declog.txt

Regarding the windchill - that anomaly occurred when I had "calculate windchill" checked, have unchecked it since then, as the VP2 is calculating it's own windchill. The first dip happened while computer was still on, rest shortly after it went off.

[steve]

I think I know what's happening now; those dashes in WL for the wind direction are the clue. I check the hiwindspeed, windspeed, and direction for the 'error' values, and if any of them give the error value, I ignore the wind data (so the previous value continues to be current). The dashes in WL suggest that when the wind is calm, the direction gets set to the error value - or something similar. I'm pretty sure this doesn't happen to me; I do see zero wind even when it's from the logger.

Anyway, I'll let you have the version with the tracing in the next half hour or so after I've tested it, so we can see exactly what values are coming back.

[steve]

Please unzip the cumulus.exe from the attachment and replace your existing file. If you also copy the backup files from this morning into your data folder before starting it up, it will reset to this morning's position and download the overnight data from the logger (and also all of today's too). Let it run for a few minutes (the diagnostic logger runs in the background and takes a while to catch up if there's a lot to log) and then attach the latest file from the diags folder (you'll probably have to zip it).

Thanks,
Steve

[DaveNZ]

The earliest back-up log was at 12:48pm today, well after the overnight data was already downloaded.
I can attach something here for you tomorrow, when I download the night's data using the new Cumulus version you have supplied me.