Correcting weather data anomolies

[sjukmidlands]

I reecently started using Cumulus. I have a techno-line (LaCrosse) WS-2300 which seemed to be downloading data sets fine when starting Cumulus, but have noticed a couple of obvious anomolies in the data the past day or two when downloaded from the weather station.

This may be the intermittent problem referred to on the main Cumulus site with extracting data via serial-USB from a La Crosse/techno-line station I suppose.

However, I've tried editing the monthly logo file to remove the extremes (40c outside temp a day or so ago, 114mph wind gists at 3.30am this morning, when it was fairly calm (and as shown either side of this data in the log file) but anomolous extremes still show up in the graphs in Cumulus and therefore also in the files on the website http://www.imagearchitect.co.uk/weather

Is there something I'm missing in editing this data to correct it as best i can?

Also, if I used a direct Serial connection rather than via USB would it likely be more reliable, or alternatively what other weather stations would you recommend which are more reliable with data extraction, without spending huge amounts of money?

Thanks!
Steve

[steve]

Is there something I'm missing in editing this data to correct it as best i can?

The daily highs and lows etc are logged in dayfile.txt, so you probably also need to edit that.

Also, if I used a direct Serial connection rather than via USB would it likely be more reliable,

I can only speak from the Cumulus point of view, but it does seem to me that most people who switch from a USB adapter to a real serial port get much better results.

or alternatively what other weather stations would you recommend which are more reliable with data extraction, without spending huge amounts of money

There are the cheap Fine Offset stations. However, these do seem prone to producing odd figures themselves from time to time; Cumulus filters out what it can. It depend how much is "huge amounts", really. I found my Oregon Scientific station to be quite reliable in terms of not producing spurious results. Note that currently Cumulus only works with the 'obsolete' Oregon stations (WMR928 etc) which don't have loggers (so you need to run them 24 hours a day to get proper results).

There's the new Vantage Vue from Davis, but you're looking at around £400 for that, with the logger/PC interface.

There are undoubtedly other stations which don't work with Cumulus which may well be good buys. The La Crosse itself possibly works a bit better with other software.

[sjukmidlands]

Thanks Steve, brilliant advice!

I'm going to try a direct Serial connection soon then, and hopefully that will fix these anomolies. Regarding another weather station, whichever I were to use, if I did change my weather station, will be one which works with Cumulus as this is the software I've decided to use, and like! I cerainly won't be moving to another software program, rest assured!

I'll try editing the day file now, of course cannot edit the entry for the current day one presumably yet.

Thanks!

Steve

[sjukmidlands]

Ok. I've edited the two data files correctly now, the month and the day file, and the main graphs all appear acceptable now, in Cumulus and on the web pages.

However the Recent Extremes on the main Cumulus screen still show the anomolies - 114MPH high wind at 03:30 today, and Maximum Temp of 40.7 at 03:41 yesterday. These are also replicated on the Today and Yesterday screens on the website of course.

Presumably this data is stored somewhere else, but in any case will 'roll off' these screens as time passes?

Incidentally, what seems to be happening is that a decimal place is added at the wrong point, hence the temp extreme should have been about 4.0 (or 4.07? or 4.7?), comparing it to temperatures either side in the data this would be about correct, and the wind speed therefore probably 11.4, not 114. This is how I have set it anyway, now, and it all looks ok.

The other entry I had to edit (and guess) was the windchill factor at the time of the anomolous wind speed!

Steve

[steve]

Presumably this data is stored somewhere else, but in any case will 'roll off' these screens as time passes?

Yes, in today.ini and yesterday.ini. The today.ini values will get transferred to today's entry in dayfile.txt, so you can either edit them now, or fix dayfile.txt tomorrow.

Incidentally, what seems to be happening is that a decimal place is added at the wrong point, hence the temp extreme should have been about 4.0 (or 4.07? or 4.7?), comparing it to temperatures either side in the data this would be about correct, and the wind speed therefore probably 11.4, not 114. This is how I have set it anyway, now, and it all looks ok.

I've seen this in other examples. The data that gets read is held in the station in 4-bit "nibbles" (normally in BCD), and each nibble is read separately. So I guess that one reason for the odd values is that the station returns the adjacent nibble each time rather than the requested one (so maybe the data is corrupted on the way out rather than on the way in), effectively multiplying it by 10.

[sjukmidlands]

Thanks yet again Steve, I've editied the ini files and all is well now

As I don't have a serial port on my new PC I'll get a Serial PCI card to add one, plus an extension cable as the one from the weather station is only 1m long, not long enough to reach the PC. Hopefully the data transfer will be a bit more reliable! Even if it makes no difference its no big expense anyway, and at least frees up a USB port.

Thanks!

Steve