increase accuracy of fine offset rain tipper

[Gina]

For an area ratio of 5 I would expect the factor to be 0.2 ie. 1/5

[tjaliwalpa]

For an area ratio of 5 I would expect the factor to be 0.2 ie. 1/5

Oops, your correct. I stuffed up. Must be the early morning as I have just woken up. In fact the area is closer to 3 times, not 5 (55cm² to 154cm²) Does that make sense. Actually, it is not 3 times, it's 0.3571428571428571 to be exact. I've just been out and remeasured it. Sorry for misleading you all.

By the way, the official data from the BOM stattion for December rain so far is 131.4mm and my station (less than a km away) is 130.6mm

[Gina]

Sorry for misleading you all.

No problem

By the way, the official data from the BOM stattion for December rain so far is 131.4mm and my station (less than a km away) is 130.6mm

Well pretty close

[hills]

Hi Subydaz, I've increased mine by 3 times as well with calibration of .35. Originally it was really accurate compared to my 3 manual rain gauges in anything but really heavy falls. So I increased the sides of the original collector bucket and since then it has improved a lot so I can only assume I was getting some overflow.

You can see how in this thread: https://cumulus.hosiene.co.uk/viewtopic.php?f=13&t=3074&

You can also see my modifications under the "Sensors" link on my website in my sig.

[MickinMoulden]

OK, so by adding a larger catchment area and using the calibration factor in Cumulus to compensate for this (eg: If I triple the catchment area of my gauge and then go under Multipliers for rainfall in Cumulus Calibration and put in 0.33) then it will be more sensitive to smaller amounts of rain and variations in the rain rate?

So far as I see it, this should give me slightly more accurate data with respect to having less likelihood of say 0.1 or 0.2mm of rain not being counted because previously it would only measure at least 0.3mm, and thereby giving me nil rain.

I take your point though on the accuracy of the device as far as measurement goes, if the original gauge is not perfectly accurate (and of course it wouldn't be as it's hardly a precision instrument) then any error will be propagated if you go changing the catchment area etc.

With respect to heavy rain, if you have a large catch area does this cause any problems like the reservoir overflowing because the bucket can't tip fast enough or any spillage through the bucket area as a result of the moving bucket itself?

Hi Subydazz, I bench tested my WH1081 after I had increased the catchment area 3 times (55cm2 to 165cm2) and found that each tip (0.1mm) was (trying to remember.....) 12 drops, and that I had no issue with wasted drops (a drop falling into the bucket that was already tipping) up to a rate of 50mm/hr. However, upto 130mm/hr you would get a drop wasted, which would be a 1/13 or 7.7% under reading. Any rate way above this you would probably get a 15.4% under reading. Out of 16 days I've been using Cumulus in December, I had 12 rain days of which 3 times I had it go over 50mm/hr (and up to 118mm/hr). I am currently happy with this trade off. I also like the fact that not only is it more sensitive and accurate (bench tested below 50mm/hr), it also records rain earlier (and sets of my voice confirmation alarm! He he ) Obviously everyone will be different. If you never get above 50mm/hr then do it. If you always get above 50mm/hr, or crikey! 130mm/hr, then you may want to think twice.
I've added a photo. My mod is in two parts. The first mod was to prevent splash out by increasing the wall height (100mm). The 2nd part (done later) was to increase the catchment area. To make the rectangle (50mm x 110mm) proportionate I just made the larger area (50mm x the square root of 3) x (110mm x the square root of 3) using laminating sheets. To make them stronger I laminated a laminating sheet for each side and hot glued them together. It's very strong and total cost was $2 for 9 sheets (the first part was only made with one sheet). I found that making the area excactly 3 times bigger, then using a Calibration multiplyer of 0.33 to give a 0.1mm reading is very neat. I have read others make mods and get each tip of the bucket giving a funny reading like 0.111mm? Weird. If your area is slightly under, you can modify the tipping bucket to make it more accurate (initially mine, unmodified, was 9% under).

Michael (Palmerston, NT, Australia)

[MickinMoulden]

[quote="hills"]Hi Subydaz, I've increased mine by 3 times as well with calibration of .35. Originally it was really accurate compared to my 3 manual rain gauges in anything but really heavy falls. So I increased the sides of the original collector bucket and since then it has improved a lot so I can only assume I was getting some overflow.quote]

G'day Hills, love your mods, particularly to the temp/hum gauge. In regards to the rain gauge mod, did you increase the sides of the tipping buckets? What I found is that adding weight to the buckets meant that it required more drops of water to overcome the extra weight on the other side of the bucket (thus making the reading read under actual). Also, with the extra weight, the buckets would tip slower, and I've found the inaccuracy comes from when the rain rate is high, drops will be wasted into a bucket that is already tipping. The fact that your mods have made it more accurate is amazing . Well done.

Michael.

[SUBYDAZZ]

So would you be able or not to modify the bucket to catch less before it tipped rather than increasing the catchment area to give the same results?

[MickinMoulden]

So would you be able or not to modify the bucket to catch less before it tipped rather than increasing the catchment area to give the same results?

Yes, reduce the tipping angle. You could throretically reduce the tipping angle so much (raising the height of where the bucket will rest can be up to 10mm each side), which would only require a few drops to tip it (instead of like 12 drops, you would get it to tip at 12/3=4 drops ((thus reducing the tip from 0.3mm to 0.1mm)). But you need to remember that it will be more susceptible to a false reading if your rain gauge will rock in winds.

I hadn't actually thought of it that way, this would solve any problem with high rain rates giving flase readings of 7% or more, however, since you are only dealing with very small measurements that there is a high possability of not being able to fine tune it and it may well consistantly give false readings. Eg. 1 drop over and your out 20%, 1 under and your out 25%.

Another thing you will have to check is that with a reduced angle by this much you would have to make certain that the drops that a falling in a FULLY falling into the bucket and not getting split by the middle of the buckets.

I would still increase the wall height of the catchment area though.

I can forsee to get this right, and I think you could, that you will need to get the measurement correct, not only to the mm, but maybe even as close to the tenth of a mm. Good Luck, perserverence is the key.

[Gina]

If you reduce the amount required for the buckets to tip you will increase the error caused by tipping and worsen performance in high rainfall conditions even more. In light rain, accuracy is limited to one drop. The only way of increasing accuracy in light rain is to increase the catchment area. Reducing the bucket tipping volume will reduce the time to detect the onset of rain but overall will reduce the measurement accuracy.

[Nin3DSFan]

I've always found that the 0.3 mm in 5 minutes can be either too high when it's not raining much, but too low when it's torrential.

For example, if the actual rain over a 10 minute period goes up like this:

0.0
0.0
0.1
0.2
0.2
0.3
0.3
0.4
0.6
0.8
0.9

A Fine Offset rain gauge would record it like this:
0.0
0.0
0.0
0.0
0.0
0.3
0.3
0.3
0.6
0.9

The rain rates for the minutes based on how Fine Offset measures rain, would be:
0
0
0
0
0
3.6
3.6
3.6
7.2
7.2
7.2

However, if the rain tips were 0.1, rather than 0.3, the rain rates would be:
0
0
1.2
2.4
2.4
3.6
3.6
3.6
4.8
7.2
7.2

In short, this then means that rain rates are 3 times more accurate, being in multiples of 1.2 mm/hr, rather than 3.6

However, when rain is very heavy, it's the timing that is the problem.

For example, in a torrential downpour, over a 10 minute period, the rain total may look like this:

0
0
0.3
0.9
1.8
3.6
6.0
6.6
6.9
6.9
7.2

Using a 5 minute average, the rain rates for the same times would look like this:
0
0
3.6
10.8
21.6
43.2
72
75.6
72
61.2
43.2

However, measuring the rain rates over 1 minute periods, it'd look like this:
0
0
18
36
54
108
144
36
18
0
18

In this example, the maximum rain rate would be more likely to be 144 mm/hr or higher, than 75.6 mm/hr.

A real life example, would be on August 24th. We had torrential rain that would have had a rate of at least 140 mm/hr. However, a FO station may only measure this as something around 72 mm/hr. Infact, the only way FO rain rates, are accurate, is when the rain rate is actually that much for 5 minutes. Therefore, if it's raining at a constant, or near-constant rate of 10.8 mm/hr, for 5 minutes, then the actual rate is 10.8 mm/hr. However, it it was only 10.8 mm/hr for 2 minutes, then 26.7 mm/hr for 1 minute, then 7.2 mm/hr for the last 2 minutes, the average rate for those 5 minutes is 12.54 mm/hr. Therefore, you'd have a maximum rate of 12.54 mm/hr, less than half of the actual maximum rate of 26.7 mm/hr.

So, there's basically no easy way to increase the accuracy.

[Gina]

I guess one way of coping with both heavy and light rainfall would be to use two rain gauges. This is a possibility I had thought of for my weather station. No problem if you're making your own weather station but not an option for comercial stations. The incidence or extremely heavy rain seems to be increasing over the years so we may well want to improve the heavy rainfall accuracy.