Fallacy of Measuring Snowfall

“How much snow did you have?”

I was asked this question often after the Nov. 29-30 winter weather event, and my answer consistently was the same: “It doesn’t matter. What does matter was the liquid equivalent, which was 1.31 inches.”

Somehow even my fellow atmospheric scientists seemed dissatisfied with that answer, which puzzles me because it is the most scientifically meaningful.

Based on the measurement attempts about which I have read and heard, the winter storm here in Norman left around 1/4 inch of freezing rain ice, followed by perhaps a half inch to inch of sleet, layered further under 4-7 inches of snow. Of course, the snow depth measurements varied greatly, not only across town but even on a small plot of land, thanks to blowing and drifting. This has been a common problem ever since snow measurements first were collected way back in the pre-electronic era.

The customary way to report snow depth has been and remains to do an average of several readings — but how many, and where? Inconsistency in sampling sizes and methods is a major detriment to climatological analysis of snowfall records; yet the detriment is immeasurable because of the lack of rigorous documentation on how each and every snow depth reading was attained.

Another longstanding problem with snow depth reporting and data is that the snow “dryness” varies greatly, such that fine powder of 10 inches melts down to a smaller amount of water than heavy, wet snow that was formed in very moist environments that barely were cold enough to sustain ice crystal growth and maintenance of those crystals to the surface. Ten inches in Aspen in February probably won’t yield the same runoff as ten inches in San Antonio in March.

The problem distills to this: Snow depth measurements are arbitrary, inconsistent, misleading and hugely unscientific way to represent fallen winter precipitation.

The solution, as is so often the case, is shockingly simple: Stop doing it! Quit measuring and recording snow depth. Instead, as with rain, collect the precipitation in a capable gauge, in an open exposure. Then melt it. The liquid equivalent is what really counts anyway for hydrological purposes. Liquid equivalent is what goes into the official precipitation records, and for good reason. Why even bother with snow measurement at all, unless you run a ski resort? It’s a waste of time and effort.

For performing melted equivalents, heated tipping-bucket gauges exist. The standard old removable-top gauges work well for this purpose when a heated gauge isn’t affordable or practical. I’ve got one of the latter, which funnel rain into an inner plastic pipe and let overflow pour into a surrounding outer pipe. For upcoming winter precip, the funnel-cap and inner pipe can be pulled in less than two seconds. All precip then can be collected in the outer pipe for later melting and measuring (by pouring back into the scaled inner pipe). This is a perfectly valid tactic because the circumference of the funnel cap and outer pipe are the same. This means the horizontal plane through which the precip falls is the same, with or without the funnel and inner pipe; therefore, the same amount of any precip will fall into the gauge either way.

The hard part may be to remember to pull the funnel and inner pipe. This must be done, because clumps of freezing and frozen precip will clog the narrow opening, causing the funnel to overflow and/or snow collected in the funnel to blow out again, forever undervaluing the measurement.

If you don’t measure precipitation, none of the above may mean much. But if you do, make it matter and melt it.


5 Responses to “Fallacy of Measuring Snowfall”

  1. Gilbert on December 16th, 2006 2:14 am


    No, no, nooooo….oh, my.

    First, yes, melting the snow to get the liquid amount is a great idea and needs to be measured well. But so is the snowfall and snow depth. Why? Snow has tremendous societal impacts, as much as or even more so than liquid precipitation. From avalanche dangers and knowing how much fell in a threatened area, to budgeting for snow removal and what type(s) of chemical(s) to use, to deciding on school/business closures, truck reroutes, generators for hospitals, and from general inconveniences to life-threatening situations in general…

    We need a good snowfall and snow depth climatological database. As of right now, both suck, as does the methodology for doing each task during a storm. The question is not if we need to do this…we need to, now, and badly. The water lets us know about flooding, groundwater availability, how well crops will start out the season, and on and on. But don’t discount the great need for accurate snowfall measurement, and snow depth measurement. Billions of dollars and many lives are on the line here.

  2. tornado on December 16th, 2006 5:52 am

    Well, there must be a question as to whether to measure snow (especially “officially”), because I’m raising that very question!

    In order to perform many of these good deeds (with a few very narrow and specialized exceptions), it is not necessary to know snow amounts to the nearest inch, rather, that there is a given range of snow depth.

    Any idiot can estimate snow depth, or even measure it in any given spot. But again: What’s the meaning of that, especially in situations where blowing and drifting make it vary wildly? Why budget for enough snowplow operation to remove, say, 10 inches, when the nearest official measurement was taken at an airport 40 miles away, and as a singular measurement in a drift (or in a relatively bare spot…again, you just don’t know). How representative is that, and how meaningful is it to any given decision such as those cited? To base “billions of dollars and many lives” (unprovable hyperbole, BTW) on so-called measurements whose reality is unknown and unreliable seems, in and of itself, quite risky and dumb to me.

    As for the very narrow exceptions, avalanche preparedness certainly is a valid one. Snow depth, age *and* crystal type is very carefully monitored by dense arrays of sensors in many avalanche prone areas, and fed into complex models which predict the threat based on those and other variables (e.g., detailed 3-d mapping of terrain slope and vegetation). That’s an entirely different animal altogether, and far from the same as you or I or the O’Hare observer “measuring” snow. If we all had such precise and intensive means of analyzing snowfall (and I mean truly, objectively determining and analyzing its characteristics in far more ways than so-called “depth”), then I would concede the point. ‘Til then, no way. The data outside such specialized micronets is garbage.

    Now if snow is going to be measured, yes, in principle I agree it should be done accurately, with extensive averaging and reporting of *all* of its salient physical characteristics — not *just* “depth.” News flash: It isn’t. How do you propose solving that problem?

  3. Gilbert on December 17th, 2006 1:34 pm

    Wow, so much stuff to shoot down… 🙂

    Snow amounts need to be measured by the nearest inch. What if we did that for rainfall? Ease of measurement is no excuse for measuring a known quantity accurately. Sure, it’s a lot harder to do, depending on the situation. An average, like it is now, should be done, but more flexible to handle situations where it is bare-ground to 10′ drifts. Our last system up here was a great example of that. I am with you on having a range reported as well, that’s a great idea…if you truly believe that a relatively close to accurate amount isn’t possible. Question: what is the correct way to measure the liquid water equivalent from snowfall? You say “melted, it was X amount, and that’s all that should be reported”. If you can’t measure the snowfall or snow depth accurately, guess what else isn’t going to be correct?

    “Any idiot can measure snow depth”? Sure, and any idiot can fall off the side of a cliff, too. Any idiot can tell you there’s a tornado, right? Measuring meteorological quantities and documenting events is not easy! Some harder than others, to be sure, but siting of a rain gauge is not a trivial exercise, and double ditto for measuring snowfall for a given location.

    Why budget enough for a 10″ snow from a station 40 miles away? You have to gauge if that is representative of your area. In the Rockies, that is not true. From Dubuque to Rockford, 40 miles will matter, but you can be sure that if a station 40 miles away from you gets a 10″ snowstorm once a year, for example, it’s safe to say you are under the gun as well. Maybe not from the same storm, but the threat is very close to being equal, unless you are near the Great Lakes. And thankfully, around here…coop stations are less than 40 miles apart. That helps. To solve THAT problem, we need more coop and snowfall reporting observers! COCORAHS is a 1st step in the right direction to deal with this problem for severe events.

    Billions of dollars and many lives lost? OK, with the 11/30/06-12/1/06 storm we had 23 dead, including one in my city, DeKalb, IL, and millions of dollars just from power lines down. Hundreds of trucks were stranded in drifts for nearly 24 hours in some cases on I-80 and I-39. That costs BIG bucks! Throw in snowplow and ice-removing costs, so high in Illinois that 2/3rds of the state was a declared a state disaster area…lost productivity, closed businesses…lost business…and that from just one storm! I’ve heard billions from weather analysts, but let me tell you I believe it. In the trenches in this department I work for, I see first hand the cost…all too high from ONE of these storms. You could say the same thing for hail reports, tornado reports, that the insurance companies base YOUR premiums on. And you’ve described “Storm Data” as an “open sewer”, to which I cannot disagree. Tell the insurance companies they’re full of crap, however, and they’ll probably laugh in your face first, and raise your rates after that! You have to use what you’re given. And you can pull a lot of signal out of the noise if you are careful, in many cases.

    The data outside of micronets is garbage? Then lets get rid of precipitation reporting altogether. Surely, ASOS reports accurate liquid and frozen precipitation? (I am out of my mind for saying this!) Surely once-per-day reports from coops is garbage, since evaporation wrecks the report? If that is what you argue, then why do it at all?

    I fully concede that measuring snowfall–and precipitation in general–is not done consistently and not necessarily well, either. That’s why NCDC takes a lot of pain to QC the data as best it can. But that does not mean that the data we have can be very useful, for planning, for saving many lives, and a ton of $$$. We’re doing it at NIU, imperfect as it is, with keeping our people off the road during bad events, and letting them know when it is safe to go back on the road, based on the snow/ice fall reported, and road reports. Our grounds crew buys enough salt to handle 40″ of snow every year, and when it is cheap to do so. With that, we never get into a pinch; if more is needed, we can get more, just at higher cost. But we save a lot of $$$ by buying it in the off-season. and knowing how much an average winter gives in terms of snowfall really helps us a lot.

    Your turn! 🙂

  4. Rich Thompson on December 22nd, 2006 2:13 pm

    Roger, you’re taking this argument to a ridiculous extreme. I agree with Gilbert – how can you trust liquid equivalent for snow melt if you can’t measure it accurately? One depends on the other, so we should be arguing for *both* or *neither*.

  5. tornado on December 24th, 2006 12:14 am

    This is an argument well worth making, and the farthest from ridiculous. I’ve stated the problem succinctly: “Snow depth measurements are arbitrary, inconsistent, misleading and hugely unscientific way to represent fallen winter precipitation.”

    Solutions, please.

    If it’s going to be done — whether with snow, rain or any other reportable variable, it should be done in a consistent, accurate, systematically calibrated and methodologically reproducible manner. More observers doing it the same stupid way — or in reality, dozens of different stupid ways — won’t help.

    It’s not the quantity that matters as much as the *quality* of reports. I want both, but if forced to choose, would rather have a few excellent ones than a lot of crappy-to-unknown quantity. As I’ve heard Gilbert himself say: garbage in, garbage out.

    Of course heavy snow has economic impact. Duh. But how much of this is to do with post-facto depth measurement as much as the physical characteristics of the precipitation accumulation itself, and a priori preparedness and prediction (not even of specific amounts, but of ranges and probabilities)? Perhaps the events wouldn’t be so costly with better planning and better forecasts, not to mention more physically meaningful measurements of snow characteristics. [More on this below.]

    Forecasts of (high absolute value) depth ranges of snow by WFO BOU clearly helped the Denver area authorities to prepare, but they still will end up spending untold hundreds of millions in that economy — maybe more — simply by virtue of having to remove the stuff.

    Surely you don’t shovel or plow 10 inches of dry snow the same way you do 7 inches of wet snow on top of 2 inches of sleet on top of an inch of ice. Yet the depth is the same! The physical characteristics of the precipitation deposit matter far more than its depth — especially when the depth reporting is such a farce.

    This brings me back to the avalanche preparedness people, who (as I’ve also stated already), have something very close to this: A scientifically meaningful diagnostic evaluation of the snow deposit’s multivariate physical and thermal characteristics which goes far beyond its so-called depth (which is crude, inconsistent and nonstandardized), or even liquid equivalent (indisputably is more meaningful hydrologically than whatever passes for “depth” from place to place, but still just one variable).

Leave a Reply

You must be logged in to post a comment.