Trace amounts of rain

[serowe]

Steve - view this as a rewquest for a reasonably 'major' modification to the way Cumulus records rain. A mod not for the sake of it but one to have it accurately report rainfall.

At present, if you have a day - such as we had in Melbourne yesterday - where you only receive very light drizzle for a short period of time, the liklihood is that your rain gauge won't record any rain (whether it is an automatic gauge or a 'more proper' cylindrical gauge. Under these conditions if you later go through or analyse your weather data it will not show the fact that any rain has fallen unless the amount is equal to or greater than the tolerance of your recording equipment.

In Australia at least, the Bureau of Meteorology have a reporting level of Trace (usually reported as Tr in the hourly reports) - so whilst no amouint has been recorded, the fact that rain (drizzle) has actually fallen and settled, the day would then be correctly reported as a 'rain day'.

How it would be done - not sure (thus my reasoning this could be a fairly large mod requirement).

In C1 there are 2 files which would probably need to be fixed - daily and monthly. The daily would need at least Tr or Trace to be reported in the rainfall amount (if more rain falls later in the day then obvikously this would be reported as such and the Tr not needed for a DAILY report).

The hourly though would need a value to be recorded to indictae a trace was seen - eg yesterday the cricket at the MCG was stopped because they had rain - the city gauge recorded 0.4 mm but a local AWS 1km from me reported zero and my gauge (both auto and cylindrical) reported zero yet the ground was completely wet.

My thought is that a negative value could possibly be used - something like -99 or -999 - a figure that couldn't be mistaken for a valid figure and stands out if the data is printed and visually studied. Obviously something like this would affect a number of aspects of Cumulus - data, screen displays, graphs etc - but programmatically it could be handled very easily.

How C2 could handle it - well not sure as I still haven't looked at it.

None of the other weather software packages, AFAIK, handle Tr amounts of rain - but for accurate recording, it is an important element to getting a true picture of what happened 'on this day'

Thoughts, comments??

[beteljuice]

The whole point of a 'TRACE of rain descriptor' is that it too small to measure !

Ergo it's a manual observation.

[serowe]

That's correct - but the official description of the day is then no longer fine (ie absence of rain).

You could have drizzle on and off all day - this makes the day 'wet' but there could easily be no rainfall recorded during that day. Which is why I am asking for a method to record this in the data without affecting rainfall totals.

I am reasonably sure that the UK met bureau will have a similar reporting method for 'trace' amounts of rain.

[WestOz]

Don't know about the Oregon WRM918 that you have serowe but I have the WMR968 and it only measures rainfall in 1mm increments, so it's possible to have 0.9mm without it regestering as having rained - a real pain in the butt. (Not the fault of Cumulus of course.)

In theory, I suppose, it could rain every day up to 0.9mm and the water evaporates before the next rainfall. So you could miss recording quite a bit of rainfall over the year.

Another hassle with the Oregon is that it only measures Barometric Pressure in 1 hPa increments.

[serowe]

This is one of the reasons I have brought this up. I did modify my rain gauge to read in 0.5mm increments but, whilst it agreed around 75% of the time with a cyclinder right next to the gauge, there were times when it was hopelessly wrong.

Another reason I am looking for a replacement station...

[tjaliwalpa]

Sorry to point out the obvious, but if the gauge will not register small amounts of rain, then Cumulus (or any other weather software) can not record the rain. The Fine Offset stations measure .3 mm of rain. I have managed to get that close to .1 mm by increasing the rain gauge collector. It should be possible to increase the collector area of your station to get your gauge and therefore Cumulus to measure smaller amounts than .5 mm.

In Australlia the BOM automated stations I monitor measure .2 mm as the smallest amounts recorded. As far as I am aware, these amounts are not recorded by BOM as a trace with the auto mated stations. Manual stations (usually read 3 times per day in my area) do have trace amounts recorded in the data, more to do with the inability of the equipment to measure small amounts, than anything else. In other words, a hint of moisture in the gauge.

One issue I have experienced with my .1 mm readings is the recording of a heavy dew as rain, or even small amounts of overnight rain which have not cause a bucket tip added to with early morning dew and causing rain to be recorded at that time. There is little I can do about this issue as I have opted to modify my gauge to bring about this.

Bob

[serowe]

Bob - I KNOW the automated stations can't read these trace amounts - if you read what I originally wrote you will see that.

This is NOT what I am asking to be recorded - amounts. Merely the fact that it DID rain but in such a small amount it couldn't be recorded.

As for the BoM not recording these trace amounts with AWS's - yes I am aware of that. I am also very aware that human populated reporting stations DO record trace amounts.

And just to set the record straight - the reaosn I know? I guess the 25 years I spent working at the Bureau in their communicaitons and satellite centre gave me just a bit of an insight as to how things work!

[DaveNZ]

This is NOT what I am asking to be recorded - amounts. Merely the fact that it DID rain but in such a small amount it couldn't be recorded.

But how would it know it did rain if it couldn't record anything? Or do you mean manual entry into Cumulus

[serowe]

Obviously - you need to edit one of two files - daily or monthyl AFTER the day. If it has recorded rainfall of any amount, there is no problem, but if nothing has been recorded in the gauge, then an edit of some kind needs to be performed BUT Cumulus needs to know about this for later re-processing.

[serowe]

One issue I have experienced with my .1 mm readings is the recording of a heavy dew as rain, or even small amounts of overnight rain which have not cause a bucket tip added to with early morning dew and causing rain to be recorded at that time. There is little I can do about this issue as I have opted to modify my gauge to bring about this.

Bob

Except that your readings are now false. Regardless of how you look at it, no amount shold be recorded if your gauge (whatever type it is) can't record it.

5 days of a Trace amount cannot be quantified other than to say '5 days have occurred where there was insufficient rain to record any reading.'

Your method is now saying that over a 1 thousand square metre area there was an amount of 100 litres on the ground - which is false as this never happened.

[tjaliwalpa]

One issue I have experienced with my .1 mm readings is the recording of a heavy dew as rain, or even small amounts of overnight rain which have not cause a bucket tip added to with early morning dew and causing rain to be recorded at that time. There is little I can do about this issue as I have opted to modify my gauge to bring about this.

Bob

Except that your readings are now false. Regardless of how you look at it, no amount shold be recorded if your gauge (whatever type it is) can't record it.

5 days of a Trace amount cannot be quantified other than to say '5 days have occurred where there was insufficient rain to record any reading.'

Your method is now saying that over a 1 thousand square metre area there was an amount of 100 litres on the ground - which is false as this never happened.

I disagree. .1 of a mm is .1 of a mm, and a trace of rain is a trace of rain. It did happen, in both instances. That is the whole point, otherwise why is anyone interested in a trace if it never happened.

Look at it another way: there will be days where minute amounts don't register as the final tip never happens or the amount in the bucket (or collector) evapourates. These never record, but they happen. Statistically, over time, such errors (and those of trace amounts and dew collected) which register will cancel (average) each other out. This will be the case for BOM auto stations as well, where .2 is measured. No doubt, heavy dew would register in these stations from time to time, particularly if it rains following the dew.

Regarding heavy dew, I suggest that manual BOM stations would also record morning dew as trace amounts.

In summary, I also know the equipment I have has its limitations with accuracy on many fronts, but I am prepared to live with that. In the end, I find an unusually high degree of correlation with our local BOM statistics when it comes to rain records, wind, humidity and pressure. Temperature is one data area I need to experiment with, particularly on cloud free days.

Bob

[serowe]

Bob - I'm not going to argue with you - read what I said - a trace of rain is NOT 0.1 mm no matter how you try and interpret it. This is NOT the way the Bureau record it - never have. The word Trace is used to solely indicate that ion this day some rain fell but it was insufficient to register a recordable amount. You CANNOT try to place a value on the trace amount which you have stated you do.

[Gina]

I agree with you Bob. If anything these tipping bucket rain gauges under record precipitation rather than the reverse, particularly in very light or "trace" amounts which may evaporate before being recorded or heavy rain when there is loss when the buckets tip. A trace of precipitation whether it be rain or dew is precipitation. Whether it evaporates from the ground or by transpiration from plants (particularly grass) is another matter. This can also be measured by instrumentation if required.

[tjaliwalpa]

Bob - I'm not going to argue with you - read what I said - a trace of rain is NOT 0.1 mm no matter how you try and interpret it. This is NOT the way the Bureau record it - never have. The word Trace is used to solely indicate that ion this day some rain fell but it was insufficient to register a recordable amount. You CANNOT try to place a value on the trace amount which you have stated you do.

No, let's no argue about it. I have always avoided the use of capitalisation in my posts for this very reason.

Respectfully, please note I am not suggesting trace amounts < .1mm be recorded as .1mm and never have. What I have said is that .1mm is .1mm of rain. I will concede BOM does not register < .2mm where as my station does register < .1mm.

[serowe]

Gina - sorry - have to disagree with you. The example I gave before is a good example of trace not bveing able to be measured by instrumentaiton whether it be automatic (tipping bucket etc) or colleciton vessel. Look at the Australian Bureau site for rainfall on Monday Dec 26 2010 and look at its distribution. Scoresby - the AWS cloest to me - recorded zero - I have video footage between the hours of 0920 and 1001 that clearly show there was precipitation that was enough to dampen the road and driveway in this area but NOT enough to record in ANY gauge - my cylindrical gauge hd some droplets around the outside and in the collector but, again, not enough to record.

No matter how you want to look at this, you CANNOT record this as any amount of rain - the word Trace is used by our Bureau to signify just this occurrence. It was in use when I first started with them in 1969and was still in use when I left in the mid 90's and is still in use today.

I am NOT asking Steve to implement anything that records a quantifiable amount - but a method or value that WILL indicate precipitation fell on such and such a date but could not be measured. This WILL have to be a manual entry of some kind BUT Cumulus needs to know about it in case it re-processes a file with this value in it and it can then be used by Cumulus to display an ACCURATE record of all rain days.

If anyone modifies a rain gauge to measure this trace amount of rain it calls into quesiton all readings from that gauge.