Introduction

Eventually I decided I should come up with a more accurate converstion factor, and then trim the significant figures. I even made a spreadsheet, and was a bit surprised that using 3.1416 for pi and 2.54 for cm/inch, and 2" for the radius of the 4" funnel, πr²h; becomes 3.1416 × (2 × 2.54)² × (1 × 2.54) = 205.926 cm³/in of rainfall. (1 cm³ is 1 milliliter (ml), a more common term.) I hadn't applied a factor for the density of water, which is 1.00 at 3.98°C. At my typical room temperature of about 20°C (68°F), water's density is 0.99823 g/ml. It's currently summer, and room temperature is higher, about 25°C, and water's density is 0.99707 g/ml. That's close to 1, but it does mean that inch of rainfall is a measurably different 205.323 grams.

Note that it's easier to weigh rainfall than try to read it on the rain gauge. From discussions on the CoCoRaHS FaceBook page, I'm convinced that a lot of people aren't reading the meniscus "properly," i.e. read the bottom of the meniscus. On days where there is more than one inch of rainfall in the gauge, it's much quicker to weigh the gross weight (gauge and rain), and subtract the weight of the gauge (tare weight) to get the weight of the rainfall (net weight). It also reduces the risk of spills while transferring water from the outer tube to inner tube.

I quickly discovered that using 205 or 206 for a conversion factor gave me rainfall amounts that were more different than simply using the 200 factor! They gave me results that were consistantly less than what I saw on the rain gauge.

I resolved to look into to this some day, but then got tied up with divorce, moving, spending too much time on FaceBook, and having a job.

I needed better equipment too. My scale came from some surplus catalog that was common before the WWW came along. I.e. it was was old. It also had 1 gram resolution. It was accurate to some 0.5%, but I needed a scale with 0.1 gram resolution. I found a $12 scale with 0.01 gram resolution and 0.03 gram accuracy. Amazing. I spent more money on calibration weights to see if the scale was accurate - it is. I had an old digital caliper that had 0.001" resolution, but it died one day, so I replaced it with one with 0.0005" resolution.

To take full advantage of this level of accuracy requires much more care in handling everything. It also allows answering questions that couldn't be touched before, e.g. what is the impact of evaporation from within a rain gauge?

Recently a new rain gauge appeared on the market from Australia, the Outback Blue (OB) gauge. Superficially it is designed similarly to Productive Alternatives's (PA) Stratus gauge that is the sole gauge used by CoCoRaHS. However, an early criticism of the gauge was that it provides rainfall amounts consistantly lower than the PA gauge. Okay, that doesn't answer my interest - "which gauge is more accurate?" It does say it's time to order one and compare both to what the design parameters are.

I had two old CoCoRaHS approved gauges, the Productive Alternatives Stratus gauge, PA for short. I bought one before CoCoRaHS reached New Hampshire, I won the other as a door prize at CoCoRaHS training in 2010. So I bought a third to have a new gauge before it outgasses a lot of whatever it is that PA gauges smell of. Dimensional stability over time can be important too.

Dimensions

• Diameter of funnel opening
  Everything the gauge catches passes the area of the cross section at the top of the funnel. I.e. They both claim to be 4" diameter circles. It turns out that when wind gives rain a horizontal velocity component, things still work well - the rain drops "see" an elliptical cross section, but they're moving faster, and the funnel catches the same number of rain drops. Secondary effects come into play with splashing on the side of the funnel and modified airflow, but for the most part a gauge that is mounted vertically will work well.

• The diameter of the inner tube.
  The inner tube squeezes the rain into a smaller space so some amount of rain needs more height to hold it. This makes it easier to read rain at the high resolution we want.

• The scale marking on the inner tube.
  At some point we need a ruler, and the spacing of the marks depends on the cross section of both funnel and inner tube.

• The last two can be combined - all that's needed is that the scale accurately marks the water volume in the inner tube.

Note that the diameter of the outer tube is immaterial, at least until you take the funnel off to catch snowfall. I won't address snowfall here.

Here are some measurements of the PA and OB gauges. For the funnel and outer tube, I measured near the top because that area is critical to getting an accurate sample. I measured further in on the inner tubes as that is not so critical and I can rest the caliper body on top of the tube. I will have to revisit the OB tubes as both are conical, not cylindrical. It might be worthwhile to verify the spacings of the marks on the OB inner tube.

Notes

• Diameters were directly measured and are the average of two or three measurments.
  I'm not completely happy with them, challenges included measuring an actual diameter instead of slightly off, measuring near the top instead of what's more convenient for the calipers, etc.

• It didn't occur to me at first to adjust for water density.
  Water has a density of 1.000 gram per ml at 3.98°C. In the section below I used water from a two liter bottle that spent hours at room temperature. It's tap water, but very clean tap water. Clean rain is essentially distilled water with a little dissolved CO2 and some cloud condensation nuclei. I couldn't find a measurement for its density on Google, but it's got to be so close to pure water as to make the exercise pointless.

• Note that ignoring density implies a 0.3% error in measurements.
  While that's barely visible when reading the scale on the inner tube, it is important overall.

• Note to self:
  For version 2, consider density (maybe at 10°C), do everything in metric.

• The OB tubes are slightly conical
  For the inner tube, that presents some challenges in checking the scale on the tube. For the outer tube, that is pretty much immaterial. It might make snow core samples easier, or maybe harder. Not a concern here.

Inner tube measurements

Measuring that amount of water is a bit of a challenge, as a water drop from an eye dropper is typically 0.05 grams so it's not accurate to use. I used this procedure:

• Dispense 20.60 to 21.50 grams of water into a moderately hydrophobic polycarbonate kitchen measuring cup.

• Transfer the excess amount by repeatedly dipping a cotton swab in the water and transferring it to a sponge until 20.59 ± 0.01 grams are left. It's hard to get that last 0.01 gram right!

• First time measurement: discard and redo.
  This allows "preloading" the measuring cup with the small droplets of water that remain between samples.

• Subsequent measurements: zero the scale before each measurement.
  This ensures each measurement will add the desired amount to the cup, and ultimately that will be what's poured from the cup.

• Pour the water into the inner tube along the side of the tube.
  Water droplets will bead on the surface of the tube, but they will be collected as the tube files. The first reading may appear a little low, but note we're not in a high resolution arena.

• Photograph the water level in the tube
  Aim the camera horizontally on a tripod, frame the view to center on the bottom of the meniscus.

  Weigh station

  Photo station

I had a problem with the PA tubes, as the 1.00" would have been higher than the drain slot in the tube. I covered the slots with a small piece of scotch tape.

While one can see water levels are between marks on the tubes, trying to make readings down to 0.001" is nearly futile. The main qualitative things to look for are:

• Are the scale markings consistently spaced?

• Does the scale really start at zero?

• Is the scale accurate?

These can be demonstrated photographically. The old Productive Alternatives tube is on the left, the Outback Blue tube is on the right. The new gauge is not included.

There is one quantitative issue, and is easy to read from the photographs.

Notes:

• Any bias in the above 0.10" steps will be repeated ten times. At least from the steady error percentage visible with the old PA tube above, it appears that any bias is very repeatable.

• The meniscus on the PA tube looks like it is the bright horizontal line, but it's actually lower. If you look at the right side of the 0.80 photo, the bright vertical bar is bent due to refraction, and it ends at the bottom of the menisicus.

• The first couple of photos of the OB tube do not have the camera quite horizontal, it's slightly looking up at the bottom of the meniscus.

• The thickness of the meniscus on the OB gauge is almost zero. I don't know if this is a function of age or different plastic composition.

• I'll be painting the raised plastic on the OB tube. The PA tube, even with the worn paint, is much easier to read.

• On the old PA tube, the meniscus goes through the center of the 0.41" and 0.82" marks. That means the gauge is reading 2.5% high.

• On the OB tube, the readings appear a tiny bit high in the 0.3" to 0.4" photos, and low on the rest, but by less than 0.01" throughout. All appear to represent an error of less than 1%.

• When the new gauge arrived, I ran its inner tube through the procedure above. The true bottom of the meniscus was still difficult to see, but one intercepted the 0.61" mark, so the new tube appears more accurate (for the 20.59 gram measurements). I expected that any change would be to shrink in all dimensions, and that would result in higher readings.

At this point, it appears the PA gauge reads high, and the while the OB gauge reads low, it's more accurate. While the claim that the OB gauge reads low is holding up, the criticism that the OB gauge reads low is not.

System measurements

Ultimately, we need to look at the performance of the full system. It would be nice to put all the rain gauges outside, trigger a 0.90" to 1.00" rain storm, and measure. Obviously, that's not an option. Instead, here I take the amount of water the actual funnel will catch in a 1.00" rain event, accounting for the density of my room temperature water. The water was added directly to the inner tube and the cotton swab technique used to get the full accuracy. This both reduces the measuring bias of making ten samples and any issues with pouring from the measuring cup to the tube.

Here are photos for each of the three tubes:

Productive Alternatives (old)

Productive Alternatives (new)

Outback Blue

In the previous section, the OB tube was more accurate but read low and the PA tubes read high. Now the relationship is the same, but both the old PA and the OB tube have a similar error, but with different signs. The new PA tube is the most accurate.

A convenient rain storm

• Transfer out a bit of water from the inner gauge to the outer so that it reads a bit less than 1.00".

• Record that, "discard" water in inner tube.

• Transfer the remaining water from the outer tube to the inner tube, using the funnel for assistance.

• Record that amount (adding it to the first measurement), "discard" water.

• Of course, I discarded water into a bowl so I could weigh it.

I wound up with 1.44" (288 grams) from the PA gauge and 1.39" (289 grams) from the OB gauge. I used my old scale that is quite accurate, but only has 1 gram resolution. From the measured funnel areas, I would expect a difference of about 0.5% (80.3 :: 80.7). It was nice to see that match. However, the PA gauge's 1.44" means that a 1" rainfall would take exactly 200 grams. The OB gauge would take 208 grams. This suggested that the OB gauge was more accurate, and closer to the 205.9 ml calculated above for the 4.00" funnel size I assumed I had.

From the better calculations, a 1" rainfall in the PA gauge should take 203 grams, a 1.5% difference, and the OB gauge should take 204 grams, a 2% difference. Those errors are a bit more than I'd like to see, but some of that comes from the 1 gram scale resolution, some comes from inaccuracy (it may read less than 1% too high), and there are opportunities for measurement error.

Future significant rainfalls will be checked with the 0.01 gram scale.

Conclusions

Apparently the early CoCoRaHS test was for a gauge with a error of ± 4% in a wide range of weather conditions. The similar design of the two gauges suggest both will pass, they certainly do in low wind conditions.

My concern that the early criticism of the OB gauge was unfair is confirmed.

The tests above are not a complete comparison, there are several more things that can be investigated, some are included below.

There are various questions and claims that can be addressed.

Claim: "the Outback Blue rain gauge is consistently measuring less precipitation than the standard CoCoRaHS rain gauge."

The quote is from https://www.cocorahs.org/Media/Docs/CoCoRaHS-OutbackBlue.pdf . The statement certainly appears to be accurate, however it is misleading and has misled.

A couple years ago I discovered there was a problem with converting between the weight of rain water the PA gauge catches and depth it represents and that the PA consistently measures higher than the math suggests. The CoCoRaHS statement is misleading in that it doesn't mention the possibility that the Outback Blue gauge is accurate and the PA gauge is not.

This claim has misled people to post 1-star reviews of the Outback Blue on Amazon.com. At the moment, there are five one-star reviews saying that CoCoRaHS has rejected the gauge. E.g. "Received official email today saying do not use this gauge for Cocorahs measurements. They have been testing this gauge and reads low. The inner scale is not in calibration." Note how the poster changed the CoCoRaHS claim to drop the reference to the standard gauge so that it unambiguously says the OB is wrong. Or "Being a member of CocoRahas [sic] I can tell you this is NOT an accurate rain gauge...."

Claim: The Outback Blue rain gauge can be used for CoCoRaHS reports.

That was on the product page for the OB gauge, but has since been removed. It was not on the packaging for the gauge. It's time to take down or tame the 1-star reviews. Note that there are people who want a rain gauge and don't want to or cannot report precip data CoCoRaHS. For example, everyone in Australia. Or people in over 100 other countries, several where appreciate the English/Metric dual scale might be appreciated.

What is CoCoRaHS's acceptance criteria for rain gauges?

The main criterium seems to recording values within ± 4% of the Standard Rain Gauge. However, the discussion the OB gauge includes references to the importance of consistency (precision) and that appears to be valued more than accuracy, how close a measurement is to it's accepted value. There is no satisfying solution to that, but in general I prefer accuracy, at least in future data. The climate record has been criticized by attempts to "correct" past data and there are many accusations by climate skeptics that the corrections have erroneously lowered past temperatures which exacerbates current warming claims. On the other hand, some record high temperatures in Hawaii were due to equipment failure, but allowed to stand even after the equipment was fixed.

Accepting a mix of PA data (too high) and OB data (too low) can't be good. Neither can accepting those data after adjustments, especially if the raw data is discarded. What should CoCoRaHS do if it expands into another country? Start its record with data known to be inaccurate? Tell Australians that they must use the PA gauge and not the OB gauge? CoCoRaHS, welcome to the controversy. There is no escape.

Can/Should precip measurements be made by weight?

The conversion factors must be chosen either to match the reading on the gauge or to map the reading on to a more accurate amount per the previous section. Similar considerations could be applied to the OB gauge, with an appropriate conversion factor to map weight to the OB scale, the PA scale, or to a most accurate scale. It was so much simpler when the "obvious" thing to do was use 205.9, the weight of a 4" diameter disk, 1" thick of 1 g/ml water.

Are we done?

Heh. Hardly. I want to look at the weight and thickness of PA and (lighter) OB gauges, and of my old and new PA gauges. I might want to trade my new gauge for a gauge that has suffered in Arizona heat and sun and measure that.

I noticed yesterday afternoon that the PA gauge had all the water in the bottom of the inner tube, but the OB gauge had condensation haze and small droplets higher in the tube. Their black plastic and funnel opening shape deserves investigation. Reading the meniscus on the PA gauge is a challenge, I have some ideas on how to make it more visible.

References

Doesken, N. J., 2005: A ten-year comparison of daily precipitation from the 4" diameter clear plastic rain gauge versus the 8" diameter metal standard rain gauge. 13th Symp. on Meteorological Observations and Instrumentation, Savannah, GA, Amer. Meteor. Soc., 2.2. [Available online at https://ams.confex.com/ams/15AppClimate/techprogram/paper_91615.htm.]

Sevruk, B., and S. Klemm, 1989: Catalogue of national standard precipitation gauges. Instruments and Observing Methods Rep. 39, WMO Tech. Doc. WMO/TD-313, 50 pp. [Available online at http://www.wmo.int/pages/prog/www/IMOP/publications/IOM-39.pdf.]

N.B. This document was written by Eric Werme, Sutton Mills NH. Any errors that could be mine are mine. CoCoRaHS was not actively involved in the initial writing.

Contact Ric Werme or return to his home page.

Written 2019 July 28, last updated 2023 March 6.