Welcome to the Cumulus Support forum.

Latest Cumulus release v1.9.4 (build 1099) - 28 November 2014
Latest Cumulus MX release v3.0.0 build 3047 - 03 February 2019. See the Wiki for download

Rainfall accuracy

Discussion specific to Fine Offset and similar rebadged weather stations
ianbatty
Posts: 3
Joined: Wed 04 Jan 2012 10:34 pm
Weather Station: WH3081
Operating System: XP SP3
Location: Harcourt, Victoria

Rainfall accuracy

Post by ianbatty » Mon 30 Jan 2012 3:22 am

Has anyone had problems with electronic gauges registering *very* different readings from cylinder gauges?

I've just bought a weather station (mounted five feet above my roof) that tells me the last 24 hours reading is 13.5 mm, while my old-fashioned cylinder gauge sitting on the ground just ten metres (30 ft) away has filled to 21.5 mm.

The rain sensor is part of the manufacturer's sensor "stack" and is mounted on its supplied pole, which is perpendicular near as I can tell.

That's almost 50% extra on the cylinder gauge (!)

Can anyone suggest a scientfic reason for the discrepancy? Do I need to get a level and be really fussy about getting the rain sensor *exactly* level E-W and N-S (i.e. in plane)?

Grateful for any advice,

Ian.

Matt.j5b
Posts: 439
Joined: Mon 28 Nov 2011 2:13 am
Weather Station: Davis VP2 DFARS+La Crosse WS2306
Operating System: Windows 10 64 Bit
Location: Ferny Grove, Brisbane, Australia
Contact:

Re: Rainfall accuracy

Post by Matt.j5b » Mon 30 Jan 2012 5:06 am

Well I have mine mounted on a slightly non level surface, and I did use a mini level to get it perfectly level and packing material at the required corners to acheive the gauge correctly level. It was out by about 10 per cent before that, but I think making it level makes is more accurate. But I have not had major rain gauge accuracy problems.
Regards, Matt of Brisbane, Australia
Ferny Grove Weather
Cumulus MX testing

serowe
Posts: 484
Joined: Tue 03 Aug 2010 6:23 am
Weather Station: WM918
Operating System: Win Server 2008 R2
Location: Ferntree Gully, VIC, Oz
Contact:

Re: Rainfall accuracy

Post by serowe » Mon 30 Jan 2012 6:50 am

Not only discrepencies between a tipping bucket gauge and cylinder but also between tipping bucket gauges located only 10 metres apaart from each other.

For instance - here over night electronic gauge recorded 3.0 mm (1mm resolution), cylinder 3.6mm and adjacent tipping bucket gauge 4.5 mm (0.5mm resolution)

One area the tipping bucket gauge WILL be unreliable is with extreme/heavy downfall rates - the bucket can become overwhelmed with the amount of water it receives and be horrendously under reporting the actual rain.

This is why *Official* rain fall records do not use electronic means - inability to handle extreme situtaion and inability to correctly report minute amounts (including the Trace amount which some <cough> software packages refuse to acknowledge exist...)
Punctuation is the difference between 'Let's eat, grandma' and 'Let's eat grandma'

User avatar
captzero
Posts: 363
Joined: Wed 03 Mar 2010 10:20 am
Weather Station: Vantage Pro2 w/ daytime FARS
Operating System: Windows 10
Location: Brisbane, Australia
Contact:

Re: Rainfall accuracy

Post by captzero » Mon 30 Jan 2012 7:31 am

serowe wrote: which some <cough> software packages refuse to acknowledge exist...
serowe.....Most times I find your replies quite informative and you are obviously knowledgeable as far as weather is concerned (I know... 20 years with BOM... etc) but as always and with just about every post you write, you have to put the knife into Cumulus in some way. The snide comments are becoming tiresome and boring.
Dan

http://www.brisbaneliveweather.com




A man with a thermometer always knows the temperature. A man with two thermometers, not so sure.

Matt.j5b
Posts: 439
Joined: Mon 28 Nov 2011 2:13 am
Weather Station: Davis VP2 DFARS+La Crosse WS2306
Operating System: Windows 10 64 Bit
Location: Ferny Grove, Brisbane, Australia
Contact:

Re: Rainfall accuracy

Post by Matt.j5b » Mon 30 Jan 2012 7:45 am

serowe wrote:Not only discrepencies between a tipping bucket gauge and cylinder but also between tipping bucket gauges located only 10 metres apaart from each other.
I agree , that why I use for my personal proper daily rain recording I have been using a manual gauge before any weather station gauge and it is spot on all the time and won't be giving on the old gauge.
captzero wrote:but as always and with just about every post you write, you have to put the knife into Cumulus in some way. The snide comments are becoming tiresome and boring.
Please keep your comments nice, like captzero said, it is boring and not useful, Cumulus is great piece of work by Steve :), it is not the fault of Cumulus that it can't measure a trace. Your comments are very helpful, no need to knock that, just keep the comments clean.
Regards, Matt of Brisbane, Australia
Ferny Grove Weather
Cumulus MX testing

serowe
Posts: 484
Joined: Tue 03 Aug 2010 6:23 am
Weather Station: WM918
Operating System: Win Server 2008 R2
Location: Ferntree Gully, VIC, Oz
Contact:

Re: Rainfall accuracy

Post by serowe » Mon 30 Jan 2012 7:50 am

Take it any way you want - the fact is that tipping bucket gauges do NOT record 'trace' amounts which is an acceptable, and official, record of rainfall. It isn't a nide comment, as you want to call it, but I didn't mention any particular software package (because it isn't confined to just one) - as the saying goes, though - if the cap fits...

Consider the scenario though - and yes, this one IS using Cumulus as an example - now that the webtag/data 'Days since Rain' has been added, the inability to record a non-recordable rain fall means that this 'data' is no longer valid. As I said, most tipping bucket gauages are either 1.0 or 0.5mm resolution - rain can, and often does, occur with less than this amount falling (we've had 3 showers in this area today). Heat and other factors combine to ensure any residual moisture in the bucket never get large enough to be recorded, thus (and given the accuracy posters here have been yelling about 'needing' for pressure and temperature) I'll stand by my (so-called) snide comment that 'some <cough> software packages refuse to acknowledge (trace falls) exist...).

Sorry if that offends - truth and facts sometimes do. And, let's face it - what's the name this thread has been given???
Punctuation is the difference between 'Let's eat, grandma' and 'Let's eat grandma'

serowe
Posts: 484
Joined: Tue 03 Aug 2010 6:23 am
Weather Station: WM918
Operating System: Win Server 2008 R2
Location: Ferntree Gully, VIC, Oz
Contact:

Re: Rainfall accuracy

Post by serowe » Mon 30 Jan 2012 7:56 am

Matt.j5b wrote:Please keep your comments nice, like captzero said, it is boring and not useful, Cumulus is great piece of work by Steve :), it is not the fault of Cumulus that it can't measure a trace. Your comments are very helpful, no need to knock that, just keep the comments clean.
Not right - Tr. is an exceptional amount and could be catered for with proper coding. As this thread *is* about 'rainfall accuracy' it is a point that does need to be made because accumulated trace amounts which, in a cylindrical gauge can easily top more than 1 or 2 mm, can be quite easily be ignored by an automated tipping bucket. As I mentioned - I already have a discrepency of 1.75mm between the cylinder and the auto bucket here since it started raining last night around midnight - primarily because the falls have been in hot conditions and each fall has been very light (most likely evaoprated between showers during the day today).
Punctuation is the difference between 'Let's eat, grandma' and 'Let's eat grandma'

wd40
Posts: 448
Joined: Thu 05 Aug 2010 10:02 pm
Weather Station: WS2080
Operating System: W7
Location: GA, USA
Contact:

Re: Rainfall accuracy

Post by wd40 » Mon 30 Jan 2012 12:57 pm

In everyday home use, trace amounts is meaningless. In terms of rain needed to grow things, anything below 0.1 inches is kind of like the perdurable drop in the bucket.

Cumulus and my FineOffset suit my work just fine.

User avatar
Tau Bootis
Posts: 125
Joined: Sun 13 Jun 2010 11:44 pm
Weather Station: WH1081
Operating System: Windows XP Pro SP3
Location: Audley Staffordshire UK
Contact:

Re: Rainfall accuracy

Post by Tau Bootis » Mon 30 Jan 2012 3:31 pm

ianbatty wrote:Can anyone suggest a scientfic reason for the discrepancy?
Yes
Accuracy of rainfall measurement is mainly affected by wind, by the height of the gauge and exposure. Wind and exposure errors can be very large, even more than 50 percent. The catch of rainfall is a function of the height of the gauge; the more open the location the greater will be the difference in catch with height.

There is plenty of official info on the net regarding this issue, it would be best if you could get your rain sensor lower down and away from any obstructions, sometimes easier said than done.
Mark

Image

b.e.wilson
Posts: 89
Joined: Tue 26 Jul 2011 11:07 pm
Weather Station: Davis Vantage VUE
Operating System: Windows 10
Location: Utah, United States
Contact:

Re: Rainfall accuracy

Post by b.e.wilson » Mon 30 Jan 2012 4:34 pm

How is "trace" detected electronically? I doubt it would be by collection; the walls of any collection device would be too "sticky" to trace amounts of rain. Is it by continuity between elements in a grid of charged wires above a salt-like substrate? Centrifugal disc/collection?

User avatar
steve
Cumulus Author
Posts: 26717
Joined: Mon 02 Jun 2008 6:49 pm
Weather Station: None
Operating System: None
Location: Vienne, France
Contact:

Re: Rainfall accuracy

Post by steve » Mon 30 Jan 2012 5:06 pm

b.e.wilson wrote:How is "trace" detected electronically?
This device uses capacitance to detect precipitation: ftp://ftp.campbellsci.com/pub/csl/outgo ... _oct08.pdf

If you had one of these and it detected rain, but it wasn't sufficient for your TBR to tip, then you could log a 'trace'. A number of automatic UK Met Office stations occasionally return 'trace', so they presumably have one of these or something similar.
Steve

User avatar
Tau Bootis
Posts: 125
Joined: Sun 13 Jun 2010 11:44 pm
Weather Station: WH1081
Operating System: Windows XP Pro SP3
Location: Audley Staffordshire UK
Contact:

Re: Rainfall accuracy

Post by Tau Bootis » Mon 30 Jan 2012 7:42 pm

How is "trace" detected electronically?
Its not, trace is recorded when there is no sign of precipitation in the gauge but it is known for certain that slight rainfall has occurred since the last raingauge reading.
Mark

Image

serowe
Posts: 484
Joined: Tue 03 Aug 2010 6:23 am
Weather Station: WM918
Operating System: Win Server 2008 R2
Location: Ferntree Gully, VIC, Oz
Contact:

Re: Rainfall accuracy

Post by serowe » Mon 30 Jan 2012 8:22 pm

b.e.wilson wrote:How is "trace" detected electronically? I doubt it would be by collection; the walls of any collection device would be too "sticky" to trace amounts of rain. Is it by continuity between elements in a grid of charged wires above a salt-like substrate? Centrifugal disc/collection?
It isn't (as Tau Bootis states). But what a recording of 'Trace' in the data does do is show that those 10 or 20 days your electronic records show as being rain free are not, in fact, totally free of precipitation - just that on some day or days, there was precipitation BUT it was too small to be recorded.

There is also another measurement used, officialy, of 'dew' which again shows that there wasn't a recordable amount of rain or precipitation BUT the gauge did contain moisture which was due to a phenomenom known as 'dew'.

If the software (ANY software) contains so-called records of 'nn rain free days' then that software should also have the ability for user intervention to the data and a method of recording 'Trace' amounts regardless of whether the author believes them or not. Makig it user intervention and providing a method (eg an amount of -999 which is so patently wrong can easily be used in data for this) that can be used if the user wants or ignored if they 'don't believe it'.
Punctuation is the difference between 'Let's eat, grandma' and 'Let's eat grandma'

User avatar
mcrossley
Posts: 5374
Joined: Thu 07 Jan 2010 9:44 pm
Weather Station: Davis VP2
Operating System: Stretch Lite rPi
Location: Wilmslow, Cheshire, UK
Contact:

Re: Rainfall accuracy

Post by mcrossley » Mon 30 Jan 2012 8:33 pm

What about the optical rain recorders which count drops falling into the device - presumably these can register a 'trace'?

User avatar
steve
Cumulus Author
Posts: 26717
Joined: Mon 02 Jun 2008 6:49 pm
Weather Station: None
Operating System: None
Location: Vienne, France
Contact:

Re: Rainfall accuracy

Post by steve » Mon 30 Jan 2012 8:47 pm

mcrossley wrote:What about the optical rain recorders which count drops falling into the device - presumably these can register a 'trace'?
Yes, because like the device I linked to, those can also detect that it's raining from only a small amount of rain, much less than the 0.2 mm that would be required to tip a typical TBR (well, OK, 0.3 mm in the case of Fine Offset stations).
Steve

Post Reply