Tony Crocker":ljg38jqk said:
I believe that my original chart is closer to the mark than the revised chart due to the container method understating both new snow and water content.
Based on our current understanding of the causes of the decreased snowfall reporting at the Mt. Mansfield stake, the amount of correction that needs to be made to the original table appears to come down to whether the snowfall deficiency is caused by 1) a portion of the snow simply not being captured by the container at all (in this instance, both the snow and it’s associated water content are excluded from measurement, there is no effect on snow density calculations, and thus no correction to the numbers in the table are needed), and 2) excessive settling of the snowfall due to the 24-hour collection interval/daytime heating/melting (if the entire snowfall deficiency was derived from this cause, then the maximal correction would be needed). With the information at hand, I’d be interested in hearing why you feel that cause #1 accounts for the lower snowfall numbers at the stake.
While it is very possible that the amount of snow captured by the gauge on Mt. Mansfield is reduced relative to what would be captured on an open, flat snow collection surface like a board (as I have observed in comparisons with my own gauge and boards), we don’t actually know how much of a factor it is for the specific gauge and location in question. I would argue that this issue is probably not the major source of the snowfall deficiency at that specific setup, and that the corrected table is a better depiction of the snowfall in the higher elevations of Mt. Mansfield for the following reasons:
1) Regardless of whether snowfall is being missed by the gauge or not, whatever does fall in the gauge is still going to settle over the course of a day. Thus, in order to pick up the 230 inches of 24-hour/daytime heating/rain-settled snowfall per season that
is reported at the stake, substantially more snow than that has to fall. How much more? 10%? 20%? 30%? Higher? The rate and extent of the settling is going to vary with the density of snow, but from personal experience, I would say that even 30% settling isn’t surprising for some of the very dry snow that we receive in Northern Vermont, and numbers like that could account for a large part of the discrepancy in snowfall reporting between the stake and the ski resort. Although I don’t specifically monitor rates of snowfall settling over 24-hour periods for my own analyses in Waterbury, using the details associated with my snowfall observations can allow for some settling estimates. Here are some observations from
a storm that started on Thursday, December 31st, 2010:
On Saturday, January 2nd, 2010 at 10:30 P.M. in the evening, my observations at the house indicated that we had just picked up 6.6 inches of 3.2% H2O snow in the previous 6 hours, the temperature was 11.7 F, and there were 19 inches of snowpack at our back yard stake. The next morning, Sunday, January 3rd, 2010 at 7:00 A.M., I made my next set of observations. The temperature was 9.5 F, an additional 5.1 inches of 5.9% H2O snow had fallen on the snowboard, and the snowpack at the stake was 21 inches.
Without any settling, the expected depth of the snowpack at the Sunday morning observation would have been about 24 inches (19 inches + 5.1 inches), but it wasn’t; a full three inches of snow had been lost in that 8.5-hour period. Based on the air temperatures recorded along with the observations, the snow was clearly not melting, and the winds were not strong enough that they were packing the snow, so the loss appears to be due to settling. The observed settling at our stake was presumably not coming from the 5.1 inches that had just fallen though, since during the 8.5 hours in which it fell, it should have settled just like the 5.1 inches that had simultaneously fallen on the snowboard. However, at our back yard stake, that new 5.1 inches of snow had accumulated on top of the 6.6 inches of snow that had fallen in the previous six hours. If one assumed that all the observed loss in potential snowpack depth came from that underlying 6.6 inches of snow that had fallen in the previous round of collection, it would represent a loss of 45.4% due to compaction.
To further distribute the settling into more of the snowpack though, we can assume that the three lost inches came from all the snow that had fallen since the start of the storm on Thursday evening, December 31st, 2010 (12.9 inches). Even that calculation puts the loss of snow at 23.3%. However, that number is probably low with regard to how much settling actually occurred for the whole storm, because with the initial rounds of snowfall we are now getting back to a period more 48 hours before the measurement, and the snow had probably done some of its own settling prior to the compression from the new snow on top. Since I have several rounds of detailed snowfall and snowpack observations from the storm, it’s possible to go back and determine the additional snowpack settling from each interval. That gets pretty tedious though, and it’s fairly easy to get a sense for the settling with a look at the snowfall and snowpack change over the course of the entire storm… or even with the inclusion of previous storms.
Back on Monday the 28th and Tuesday the 29th of December, we’d also had another 9.1 inches of snow from
a different storm, so one could also think about including that into the settling calculations. But, the farther back in time and deeper into the snowpack one goes, the more the snow will have already settled, and the less and less it plays into the settling seen in the most recent snowfall. The settling process is clearly dynamic, but if one were to try to determine settling for those specific 6.6 inches from the storm above, the amount of settling is presumably going to fall somewhere in the range of percentages that I mentioned. Add to that the settling that the previous rounds of snow had already undergone before new snow had fallen on top, and one can at least get a sense of what might be going on over the course of 24 hours with these types of dry snow.
To provide some perspective on the overall settling during a longer timeframe, let’s look at the change in snowpack over the period. On the morning of Monday, December 28th, before either of those two storms came in, the snowpack at the back yard stake was at 7 inches. Since there had been some warm weather near the Christmas Holiday, the surface of that snow should have been pretty solid and not subject to much compression, so it’s a convenient place to begin an analysis of snow settling. In the ensuing several days, 27.1 inches of snow fell at the site, and the snowpack at our stake wound up at 21 inches. That represents a 14-inch increase in snowpack from 27.1 inches of snowfall. Roughly half (13.1 inches, 48.5%) of the “potential” snow depth from my six to twelve hour snowfall observations was lost in those several days.
And, let’s not forget that the settling taking place during that period was happening in a consistently cold and calm location, and that any heating, rain, or winds would further increase the settling. Even snow that is settling relatively quickly is still doing so on a timescale that people aren’t going to perceive, so I don’t think people really appreciate just how much settling goes on with the very dry snow that we get in the upslope regions. I’d argue that the settling that takes place in fallen snow of this type over the course of 24 hours, especially with other insults to the snowpack thrown in, can be quite substantial.
2) Another argument I’d make to suggest that the revised table is actually more representative of reality than the original one, is something I’ve mentioned previously in this thread. This point is based on the typical snow density used by the NWS and other forecasters for the Northeast U.S (1:10 – 1:12 H2O:snow, or 8.3% - 10.0% H2O). With all the dry upslope snow that Mt. Mansfield receives on top of the standard synoptic 8-10% H2O snow, one would expect the overall density of the snowfall to be below this range (less than 8% H2O), not at the top of it (~10% H2O) as the original table suggests.
3) Finally, I decided to calculate the average density of the snow that fell at my own house last season, because I was pretty sure that even at our spot down low in the valley, the mean snowfall density was below the 8-10% H2O range. Using the amount of snowfall and associated liquid for each storm, and weighting by the amount of snow, the calculation revealed an average snowfall density of 7.5% H2O for my location in the 2009-2010 season. In this calculation I was able to incorporate 121.5 (95.1%) of the 127.7 inches we received last season at the house. For the remaining 6.2 inches (4.9%) that fell, the snowfall either occurred outside the November through April timeframe, or I lacked all the liquid data to perform the calculation. The storms and their associated data that were used for the calculations are shown in the table below to provide an overview of the type of events and densities that are involved:
The 7.5% H2O number is the average (mean) value for the snowfall density, and it may not coincide exactly with the peak (mode)-style value that one would pull out of tables made up the way Tony has them, but the values should be close unless the distribution of snowfall density is very strange.
I would also argue that the mean density of the snow that falls (or at least the density of the snow that is measured) up on Mt. Mansfield at the stake is even lower. We’ve already discussed the relationship between snow density and elevation in this thread, and I think Powderfreak did a nice job of explaining why there shouldn’t be much of a difference in snow density at various elevations as long as all locations are under consistently sub-freezing conditions. The snow is created up in the atmosphere with a certain structure, and it’s that structure that determines the density to which it packs. Whether the snow finally settles at around 4,000’ up on Mt. Mansfield, or at around 500’ in my back yard, as long as other factors like wind or above freezing temperatures aren’t affecting the deposition, the snowfall should all have the same density. The problem is that above freezing temperatures do occur, even in the Northeast where winter temperatures are typically below freezing all the way to the valley floors. I’m sure everyone can relate to the experience of increasing snowfall density with decreasing elevation at ski areas - how many times does Whistler experience storms with powdery snow in the alpine and slush or rain at the base areas each season. I don’t think anyone believes that the same snow density is going to be recorded at both elevations. A perfect local example can be seen from this past season at Mt. Mansfield.
On Thursday, February 25th, our family skied with Stephen and his children at Stowe, and I spoke specifically about the density of the snow with respect to elevation in my report from the day:
“The uppermost elevations were in the clouds, so visibility was low up there, but those areas also had the driest snow. The middle elevations were sort of that sweet spot where visibility was up, and the snow was still nice. The lower ½ to ¼ of the mountain had great visibility, but the snow was fairly wet, so off piste skiing wasn’t quite as easy as it is with our more typical Vermont fluff.”
It didn’t take a rigorous analysis of the fallen snow to know that the density varied quite a bit over the 2,000’+ elevation range that we were skiing. The precipitation we were seeing at the time told the same story, with reasonably dry snow falling on the upper part of Mansfield, and combinations of rain and wet flakes coming down at the base elevations (~1,500’). At the house, another 1,000’ below that, the precipitation had been entirely liquid as far as I could tell, so I have a hard time seeing how the density of the measured snowfall through that 3,000’+ elevation range would not show substantial differences. That snow may have been all the same when it formed up in the clouds, but due to surface temperatures it did not settle with the same density at various elevations down below. One can look at my table above and see the high density of the Winooski Valley snow associated with those storms at the end of February; there was even some rain and very dense snow in the second storm that produced an overall density of 20.2% H2O. Meanwhile, up at the Mt. Mansfield stake, the precipitation was all snow through those storms. With Mt. Mansfield being both north of our location and at a higher elevation, barring the effects of wind etc., it would have to be an extremely rare event indeed that would ever produce snow on the upper elevations of Mt. Mansfield that was denser than what we were simultaneously picking up at the house.
Based on the above data and discussion, with the caveat that average snowfall density may not exactly mirror the peak value obtained from a table like Tony has produced, it’s hard to see how the original table is a very realistic representation of Mt. Mansfield’s snow density. Even the snowfall data obtained from my house for the 2009-2010 season, which include a substantial amount of unusually wet snow that fell at the end of February, reveal a mean snowfall density of 7.5% H2O. So, for Mt. Mansfield’s average snowfall density for November through April to be at 7% H2O or even below that value should not be too surprising. Indeed, for the Mt. Mansfield gauge associated with the stake to still be measuring 230 inches of snowfall per season after accounting for both a failure to collect a substantial portion of the snow as you suspect above,
and 24 hours of snow settling each day, then the resort observing a 333” snowfall average at that elevation is not too surprising.
