Something that I have noticed over the last several years is a disconnect between the season-to-date snowfall and the snow depth for Fairbanks. Reader Eric also commented in the previous blog post about the normal snow depth and the below normal snowfall. The chart below shows the snow depth surplus/deficit and the season-to-date snowfall surplus/deficit in inches (y-axis). Including this winter, 5 of the last 6 winters show a paradoxical inverse correlation between snowfall and snow depth. This is somewhat puzzling. Is the snow "fluffier"; i.e., more dendrites? Is there less wind to compact the snow? Unfortunately there are no snow water equivalent measurements after 2001.
Note: the normal snow depth value is from the 1981-2010 NCDC climate normal database. The value used is the daily 50th percentile. When there is less than a 50% chance that the 50th percentile snow depth exists (a coincidence) the normal snow depth is listed as 0". That's why it goes from 8" on May 2nd to 0" on May 3rd. Only days where a published normal snow depth were used in this analysis (9/26 to 5/2 each year).
Brian,
ReplyDeleteI believe part of this comes from the median being used for snow depth, but the mean for accumulated snow totals. The snow totals are positively skewed, so the mean is higher than the median and "below-normal" snowfall will be observed more often than not. For example, the 1981-2010 mean snow total through November 30 is 25.8", but the median is 22.0". This year's total of 19.0" is not as far behind the median.
However, the fact remains that the November 30 snow depth of 12" is above both the mean of 11.1" and the median of 10". So it seems there is still a paradox to be explained.
Thank you Richard. I had not delved into the NCDC methodology for calculating the normal daily snow depth. As you suggested, mean vs. median will make a difference. I added a second chart to the blog post that has an additional line showing the difference between the measured snow depth and the 1981-2010 daily average snow depth. It makes the chart a little more difficult to read. Sorry about that. The magnitude of the paradox is reduced but it remains nonetheless.
DeleteI wonder about the utility of such measurements? Snow falls, settles, sublimates, blows around, whatever. More here, less there, just like local temps dependent on location, observation methods, and equipment.
ReplyDeleteWhere and how do they measure this snow stuff?
Gary
Here's the official protocol document Gary: http://www.nws.noaa.gov/om/coop/reference/Snow_Measurement_Guidelines.pdf
DeleteYup... found this general link also: http://www.crh.noaa.gov/jkl/?n=snow_measurement
DeleteInteresting. Seems like water equivalent would be critical in the analyses.
Gary
Brian,
ReplyDeleteVery good. One other thought that just occurred to me... the snow total values (normals and observed) become considerably greater than the snow depth values by mid winter, simply because of compaction and melting. So for example at the end of January the mean (median) snow total is 48.0" (42.2") whereas the mean (median) snow depth is 18.4" (16.0"). Therefore deviations from normal for snow total will be much larger in magnitude than for snow depth, i.e. the standard deviations are much different. This must partly explain why the snow deficits are so large in comparison to the snow depth deficits. I suppose the most appropriate comparison would be to show the evolution of the percentiles for each distribution, but this would be a lot of work.
Neverthless, there still seems to be a bias for relatively high snow depths in recent years - and this is the really interesting question that I'll have to keep thinking about.