Cumulus Support

Hi,

Is there a chance of having a option to change the 'Rainfall since 00:00" to "Rainfall since 09:00" if you have the logrollover set to 9am?

Are you talking about in the web pages ?

If so ...

<#rfall> = Rain so far (Station) 'Today'

<#rmidnight> = Rain since ... midnight

http://wiki.sandaysoft.com/a/Webtags

beteljuice wrote:Are you talking about in the web pages ?

No, in Cumulus itself

If you have the rollover set to 0900, then 'rainfall today' is the rain since 0900. All of the 'today' figures start at 0900 for those with an 0900 rollover. The 'rain since midnight' is provided as an addition for those with 0900 rollovers who also want rain since midnight.

Are there any issues with changing to 9:00am rollover from 00:00 rollover after the station has been up for some time?

hills wrote:Are there any issues with changing to 9:00am rollover from 00:00 rollover after the station has been up for some time?

Cancel that, I found this:

Steve wrote: You'll get inconsistencies. Cumulus stores the data according to your end of day setting, e.g. the dayfile entries are created at the rollover time for the previous 24 hours, so if you change your rollover time, your previous data will now be assumed to represent 0900-0900. Similarly, when you're using an 0900 rollover, the first 9 hours on the first of the month are actually in the previous month's log file, because they 'belong' to that month.

It's not as bad as trying to change your units, which completely invalidates your data, but it makes your previous data "not quite right". As with the units settings, the intention is that Cumulus 2 won't have this issue.

So I assume this means today's data (if changed today) will contain 33 hours of data, not 24, ie midnight until 9:00am the following day. If so that would just be a once off hit of "not quite righteousness", and then it will all be spot on after that?

hills wrote:So I assume this means today's data (if changed today) will contain 33 hours of data, not 24, ie midnight until 9:00am the following day. If so that would just be a once off hit of "not quite righteousness", and then it will all be spot on after that?

It would mean that your data from before today would not be quite consistent with the data from today onwards in respect of things like rain totals, and the timestamps of highs and lows (e.g. where a minimum temperature occurred between midnight and 0900 - the usual case - it will have been timestamped a day later than it would have been under an 0900-0900 regime). It's perhaps a subtle difference, but it would matter if you were keeping strict records and comparing with other stations. But I don't think Cumulus actually complies with Australian rules for timestamping anyway, so this is all perhaps rather academic as far as you're concerned.

It would probably be sufficient simply to state somewhere that figures before a certain date were midnight to midnight.

ok thanks. I've converted over so now I'll be in sync with the BOM sites.

hills wrote:ok thanks. I've converted over so now I'll be in sync with the BOM sites.

umm, not quite, you will be one day out with rain and everything else. Cumulus records the data in the wrong day cf. the BOM in Oz.

Ah ok, so if at 9:00am tomorrow morning my station has recorded 20mm, it will record it against tomorrow's date (the 28th) whereas BOM will record it against the 27th, is that correct?

Even so that's not really an issue for me as this is the site I want to be in sync with: http://www.bom.gov.au/sa/flood/adelaide_metro.shtml

If I view the site at 3:00pm tomorrow, both it and my station will show the totals since 9:00am tomorrow morning, won't they?

hills wrote:Ah ok, so if at 9:00am tomorrow morning my station has recorded 20mm, it will record it against tomorrow's date (the 28th) whereas BOM will record it against the 27th, is that correct?

I think it's the other way round; Cumulus counts the period from midnight to 0900 as part of the previous day, as per UK Met Office practice, which is logical since 15 of the hours were from that day and only 9 from the following day. I'm not even sure that the BOM credits all of the data ending at 0900 to the same day; or maybe I'm thinking of another country.

If it's just a case of using 'today' instead of 'yesterday' for everything, that's not too hard to fix.

Yes, its the other way around. It isnt generally a huge problem, other than messing up monthly and annual rainfall figures (or temp records) if it rains at the ends of each period

Ok thanks, It sounds to me as if Cumulus is doing it the best way then!

The bom only show rain since 9am in there observations all other max & mins are since midnight, but for official archive data the max & min temp recorded is between 9am & 9am dated the start of the 9am day. A bit of a mixed bag if you ask me.

http://www.bom.gov.au/catalogue/observa ... ions.shtml

Thanks for that Gerard. At least now I'm in sync with you and Gordon!

Do you keep your rollover time at 9:00am all year or switch to 10:00am during daylight savings?