Monthly patterns in Scottish snow (1946-2006)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
There’s been a bit of a flurry on Twitter recently about April snow in the UK (see tweet below). My opinion is there’s often UK snow during April, but it happens to be where the population isn’t (e.g. Scottish Highlands) and the UK media is quite parochial in their reporting.
I was enjoying these tweets from sunny Portugal when they came out, but I did promise to delve into the SSGB dataset to see what patterns and trends appeared from April snow. I’ve been meaning to look at the variation in monthly snow cover for a while now; so writing some scripts to look at April, it didn’t take much of an extension to look at the other snow months too.
My plan for presenting these data was based on a plot that didn’t quite make it into my thesis. Which shows year on the x axis and elevation on the y axis; the plot is then comprised of cells, the colour of which indicates the median duration of snow cover from all reporting stations. That plot appears below. We can see some darker areas every 10 years or so, corresponding to a longer duration of snow cover. However, a sixty year record is too short to make a statement about periodicity. A key point to note is, from 1992 onwards the SSGB data collection was being wound down and finally ended in the summer of 2007; so there are a decreasing number of stations making observations, which may impact on the duration of snow observed.
The number of stations observing all six months between Nov and Apr are shown below. You can see several periods where few or no stations reporting all six months, these are reflected in the plot above where missing data are shown as white space. These will obviously have some impact on the median value of snow cover duration.
The 124 stations, which in some years recorded all months between Nov and Apr, are shown on the following map (produced in QGIS). Those stations which are darkest coloured will contribute the greatest continuity to the results presented in this post.
My next step was to plot snow cover duration for each month (Nov to Apr). There was a little bit of code involved to process data for the plot (see git repo), but I think the overall result is a clear presentation of the difference in snow cover duration between different winter months. Certain years, like Jan and Feb 1962, leap out as having a great deal of snow cover. While the early naughties show very little snow cover each winter.
Finally, here are the mean days of snow cover for each elevation in each month (essentially a compression of the above plot):
Elevation (m) | Nov | Dec | Jan | Feb | Mar | Apr |
1200 | 21 | 25 | 25 | 24 | 26 | 25 |
1050 | 19 | 23 | 26 | 25 | 25 | 21 |
900 | 15 | 22 | 25 | 23 | 23 | 18 |
750 | 11 | 18 | 21 | 20 | 18 | 12 |
600 | 8 | 14 | 18 | 17 | 14 | 8 |
450 | 6 | 11 | 15 | 15 | 10 | 5 |
300 | 5 | 8 | 11 | 11 | 7 | 3 |
150 | 3 | 6 | 8 | 7 | 5 | 2 |
0 | 3 | 3 | 5 | 4 | 3 | 2 |
Note that these results are dependent on which SSGB stations reported in each year, and those years with fewer stations are going to be biased towards the results that were recorded. At some point I hope to put these results into a journal article, probably in Weather. I’d love to expand on this to look at temporal trends and see if the results are spatially organised. A closing remark would be that we (the UK) don’t currently have a wide spread systematic collection of mountain snow cover data (the closest is probably by the SAIS). Most Met Office stations are unstaffed and at low elevations. As snow gauges are expensive and a lot of the UK isn’t snowy they’re not generally used. I suspect as the SSGB was phased out there was the belief that remote sensing (satellite) data would provide the same (or better) information. Satellite snow cover data is collected by a number of instruments; unfortunately the methods don’t work well in cloudy skies or with wet snow. So, we’re left with a paucity of snow observations in the places where the most snow falls/accumlates and it would be most useful to know of snow pack extents. I suspect the answer to future snow data collection lies with remote sensing, but supported by crowd sourced data from people like those the mountaineering community.
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.