Some questions about how data is collected

[markrudling]

Hi Guys
I have some questions regarding how data is collected:
Realtime.txt – Is the data here at the exact time that the data was sampled, or an average over the time it is scheduled (Every 30 seconds etc)
DataLog – Same as above? I presume this is averaged over the time period?
The reason I ask is that I want my primary graphs to be generated using jpgraphs. Ideally I want to have as much raw data available, then use the graphs to calculate the averages based on my specifications.
I’m using a WS2300 wired. This pulls data every 8 seconds. Any suggestions on how to get the most useful data? My primary focus is on wind speeds and direction.

Kind Regards
Mark

[steve]

All of the data are 'snapshots' i.e. the value of those items at the time of log. Of course, some of the items are themselves averages over various periods, but the same applies to them. In other words, the log interval has no bearing on the data values.

[markrudling]

Perfect,

Is it safe to say, that if the weather station samples every 8 seconds, to get the most accurage logs, you should sample and log data every 8 seconds?

Thanks
Mark

[steve]

Is it safe to say, that if the weather station samples every 8 seconds, to get the most accurage logs, you should sample and log data every 8 seconds?

I assume you mean the realtime interval - so yes, that sounds reasonable. But bear in mind that Cumulus attempts to read the data every five seconds; actually, it reads the data, waits 5 seconds, reads the data again, etc. I don't know how long it actually takes to read the data; I suspect it varies, and then there's also the issue that the La Crosse interface is very poor and the station doesn't actually like talking to the PC, so many retries are often necessary.

[markrudling]

Thanks once again Steve for the support.

As the samples from the weather station are done every 8 seconds, I will see set the realtimeupdate to 8 seconds and expect duplicated data from time to time. Hopefully with such a high number of samples, there will be enough unique rows of data to have a good reading.

Thanks again
Mark