Using GIS techniques to analyse and model the topographical environment and dependencies of long-lasting snowpatch locations in the Scottish mountains

The author at the An Cùl Choire snowpatch on Aonach Beag in October 2010

One of my pet interests is the study of long-lasting (and sometimes ‘perennial’) snowpatches in the Scottish mountains. I have written many previous postings on my blog on this subject (see a list of these on my website here).

The question of what factors affect the longevity of snowpatches in the Scottish mountains through the summer and autumn seasons, and potentially until they are covered permanently by the snow of a subsequent winter (making them ‘perennial’), is one which has been discussed at some length in the relevant academic and scientific literature since the early 20th century (see a complete list of these references on my website here). Simply put, these factors are many and varied, but include:

• short-term prevailing meteorological (‘weather’) trends over the preceding months, particularly the winter months, but also the spring and summer months – these trends are: precipitation (volume of accumulated snowfall), average air temperatures, freeze/thaw cycles (which consolidate any snowpack making it more resistant to melting), effective hours of sunlight and cloud cover, and prevailing wind direction and strength [13]
• latitude and altitude [16], [17]
• local or adjacent topography (e.g. slope angle, slope aspect, upwind snow-gathering plateau areas or ‘fetch’ and sheltered depressions or hollows) [13]
• vegetation – this is largely due to high altitude tree cover and does not apply in upland areas of the British Isles, but is a factor in places like the European Alps or the North American Rocky Mountains [4], [6]
• highly localised factors such as snow avalanches in the winter months, rockfall, streams, ground composition (bare rock and soil insulate the snow differently) and snow surface coverings (such as windblown grass and soil) [1]
• human factors; it has been speculated that snowholes dug into snowbanks during the winter months for winter mountaineering skills courses may accelerate melting of the snow; this has been suggested for the Ciste Mhearad and Coire Domhain snowpatches on the Cairngorm plateau – also the ski fences erected in Coire Cas in the early 70s for the Cairn Gorm ski centre may have reduced the depth of snow in areas that held long-lasting snowpatches previous to the construction of the ski slopes in the area [1]

The effect of long-term climate trends on long-lasting snowpatches in the Scottish mountains is currently a matter of much debate. Whether measurements of these snowpatches can be used as reliable indicators of global climate change is uncertain and beyond the scope of what I am discussing in this blog post (see a BBC website news article from June 2008 here highlighting some SNH research into this). Adam Watson has been collecting data on Scottish snowpatches since the 1930s, and this rich dataset has recently (2011) been published in full along with Adam’s statistical analyses of this data (and data from other sources) and conclusions about trends [1]. This data provides the most useful source of any potential future analysis correlating the temporal and spatial distribution of snowpatches in the Scottish mountains with historic climate variation (at least over the past century or so).

The second of these factors, latitude and altitude, I discussed in some detail in my previous blog posting ‘The Scottish mountains: on the glacial ‘knife-edge’‘).

The third of these factors is one which particularly interests me, and which I believe has not been studied in great detail before in the context of snowpatch survival; most scientific research related to this subject has involved the study of glaciers beyond Scotland and also of glaciers in Scotland during the last ice-age. Much of this research has also focused on the first of the factors listed above (meteorological) and wind-drifted snow, rather than on purely topographical factors [8], [9], [10], [13].

I believe that there is scope in using GIS (Geographical Information Systems or Science)-based topographic analysis techniques to build on this research to study and analyse the topography of long-lasting contemporary (i.e. since modern useful recording and measurements began in the 20th century) snowpatch areas in the Scottish mountains, which may well prove to be a fruitful area of future research and provide useful results.

Many primary topographical attributes (such as altitude, slope aspect, slope angle, upslope slope, upslope height, upslope area, profile curvature and plan curvature) can be obtained and derived from a DEM (Digital Elevation Model), which are relevant to the locations of long-lasting snowpatches [14].

The dependency of a particular, significant and recorded snowpatch location on these attributes can be modelled using GIS and statistical software to discover any significant correlations. Hence the observed and measured spatial distribution of long-lasting snowpatches can perhaps be explained, modelled (using regression analysis) and predicted to some degree [6], [7], [13].

Unlike meteorological factors, topographical factors are relatively constant through time periods in their influence on the spatial distribution of snow, and hence are simpler to model.

In particular, GIS techniques can be used to model the effect of sunlight on snowpatch areas and calculate any dependencies on local topographical attributes affecting that sunlight (and here adjacency and proximity must be defined according to sensible criteria; a maximum distance of 10km from the location under analysis is used in [5]). This can be done by using GIS software such as ArcGIS, which can calculate the total amount of annual solar radiation (or more precisely, irradiance or insolation) on a particular location [12].

Example direct solar radiation viewshed map for a single 24-hour period; purple represents area receiving 7.68 hours of sunlight (maximum), blue represents 3.6 hours of sunlight (minimum). Image from http://freegeographytools.com/

Given DEM data for a given location (including primarily latitude and altitude data), ArcGIS can calculate the total amount of time direct solar radiation is incident on that point over an annual season cycle, taking account of shading from local topography (within a specified distance), and also taking account of indirect radiation (i.e. ‘diffuse’ scattered sunlight and reflections from surrounding topography) to produce a cumulative intervisibility ‘viewshed‘ map for that location [18], [19], see example to the right.

This technique requires plenty of computer processing power and access to high-quality, accurate and high-resolution DEM data of the area being analysed. The time interval and maximum distance from the location (or ‘viewpoint’) chosen to analyse the data has an important effect; smaller time intervals and larger distances give more accurate results, but have a proportionate influence on computational complexity and processing speed. A time interval of 30 minutes giving a reasonable level of accuracy is specified in [12].

LiDAR data represents the ideal DEM data for this technique, although availability and coverage of this data for the area of the Scottish mountains is something that would require investigation. OS DEM data (Land-Form Profile) is available at a resolution of 10m for the whole of Great Britain, which may be of a high enough resolution to obtain useful results. Unfortunately the higher resolution OS DEM data (Land-Form PROFILE Plus), which is derived from LiDAR data, is only currently available for selected parts of England and Wales.

This technique is already used in other contexts such as urban planning, solar energy planning and agriculture but it is still relatively under-utilised in the analysis of snow cover in upland environments, and has not been applied to mountain environments in the British Isles (but see [4], [9] for examples from Glacier National Park, Montana, USA, [7] for an example from Idaho, USA, and [2], [6] for examples from the Swiss and Italian Alps respectively, which use GIS techniques to analyse and model the effect of topography on snow cover and snow depth in a mountain environment).

One long-lasting snowpatch which demonstrates the potential utility of this technique is the An Cùl Choire snowpatch below the north face of Aonach Beag in Lochaber (see my website page about this snowpatch here). The An Cùl Choire snowpatch is unusual in being at such a low altitude (about 920m, 300m lower than the summit of Aonach Beag), and strangely, having gone largely unnoticed and unreported until the last decade despite being visible from the main walking route between Aonach Mòr and Aonach Beag [1, and the author's personal observations]. This may well be because the snowpatch looks uncannily similar to large areas of quartz outcrops in the corrie, especially from a distance, and casual walkers may well have seen it many times but not realised that it was a long-lasting patch of snow.

An Cùl Choire has a distinct topography which may well contribute to the survival of snow within it well into autumn and often into the winter of the subsequent year – the southern wall of the corrie rises up so steeply that any snow lying at the base of this wall to the north sees very little direct sunlight throughout the entire year, even in midsummer. How much annual radiation it actually receives can be calculated using the GIS techniques described above.

I had wondered if the An Cùl Choire snowpatch was in a location that never received any direct sunlight even in midsummer, although a photograph of Aonach Beag on Flickr here taken on July 24th 2011, proves that it does, and must receive sunlight at least 1 month either side of the midsummer solstice.

Another location which has a similar topography to An Cùl Choire is Garbh Choire Mòr, south of the summit of Braeriach in the Cairngom mountains at an altitude of about 1140m. This location contains the most persistent snowpatch in the British Isles, called the ‘Sphinx’ patch, which is surrounded by steep cliffs to the south, west and north. This snowpatch has only melted completely five times since recorded observations began at the end of the 18th century, in 1933, 1959, 1996, 2003 and 2006 [3].

These two long-lasting snowpatch locations are good candidate targets for analysis with the GIS techniques outlined here. Other candidate locations are on the north face of Ben Nevis, and there are many more throughout the Scottish mountains although the majority of the most persistent snowpatch locations are in Lochaber and the Cairngorm mountains [5].

One advantage of using the GIS techniques outlined here to calculate the amount of total annual solar radiation a location gets is that it provides quantitative results that can be used for rigorous scientific and statistical analysis, rather than qualitative results obtained from manual observations and photographic records.

Another advantage of using these GIS techniques (which ultimately rely on DEM datasets derived from data obtained using primarily remote-sensing methods such as airborne LiDAR and satellite-based radar) is that these locations are often inaccessible or obscured from view (due to difficult terrain or weather conditions) and hence not amenable to providing a consistent and long timebase duration of data gathered by on-site manual observations in the field, a necessary requirement for calculating total annual solar radiation.

The combination of remote sensing and GIS techniques is potentially a very productive one for analysing the mountain environment [4], [9], [11]; however the availability and usefulness of passive multispectral natural radiation remote sensing imagery for differentiating, classifying and analysing snowpatches [15] in the Scottish mountains is uncertain. For the image data to be useful for snow cover classification in the case of Scottish mountain snowpatches, it must have high enough spatial resolution (so that snowpatch features of the order of 10m in scale can be differentiated) and temporal resolution (to cover annual, seasonal and monthly variation), and also adequate high-quality (i.e. no cloud cover) coverage of the areas under analysis. Satellite-based remote sensing data such as Landsat and ASTER data may not meet these requirements (the CEH Land Cover Map of Britain, derived from Landsat Thematic Mapper imagery, does not include a classification for snow cover, and has a resolution of 25m). Thus primary empirical data gathered in the field and from secondary sources such as DEM datasets (derived from radar-based active remote sensing technologies such as LiDAR) may be the only practical way to analyse and model Scottish snowpatches quantitatively.

Gathering reliable and scientifically useful data of this sort on field trips into the Scottish mountains is difficult, as shown in an example given by Adam Watson (author of [1] and joint author of [3], [5], [13]), the acknowledged authority on recording data on snowpatches in the Scottish mountains (quote taken from a posting to the Winterhighland website discussion forum on 28th July 2011):

“I have some visual observations for Garbh Choire Mor in a few months in late summer and autumn, which show that the Pinnacles patch gets sunshine later in the day than the Sphinx one. The observations reveal when the sun passes from Sphinx and then a while later from Pinnacles. Of course they refer only to those dates [my emphasis]….The extent to which the perennial patches are shaded by cliffs has been exaggerated in past writings. It makes a better story to emphasise the gloomy cliffs and shaded patches! I always thought the perennial patch at An Cul Choire of Aonach Beag must be in full sunshine for hours in the morning during high summer, when the sun rises approximately in the north-east. I think this would apply also to the Observatory and even more so the Point Five patches. An interesting exception might be the patches right into the gully of Gardyloo, but of course they never attain a great size because of the narrowness of the gully and the steepness. Nevertheless they are remarkably persistent despite their small size and depth.”

There are two factors which should be taken into consideration in drawing any conclusions from the use of these GIS techniques to analyse the effect of solar radiation though:

• there is no allowance for the variation of solar radiation by atmospheric conditions (i.e. clouds, haze and fog)
• solar radiation is not the primary factor influencing snowpatch longevity, despite what one might think intuitively; current research suggests prevailing wind direction in the preceding winter months is more strongly related to longevity. Solar radiation is also almost certainly not the primary topographically-related factor; this is probably snow accumulation caused by wind-drifted snow which is affected by topographical factors such as a large plateau adjacent to a downwind depression [10], [13]

References:

[1] Watson A. 2011. A snow book, northern Scotland. Paragon Publishing.

[2] Schmidt S. 2010. Snow Cover Duration in Relation to Topography in the Loetschental, Switzerland. Landform – Structure, Evolution, Process Control: Proceedings of the International Symposium on Landform organised by the Research Training Group 437. 151-164.

[3] Watson A, Duncan D, Pottie J. 2007. No Scottish snow survives until winter 2006/07. Weather 62. 71-73.

[4] Geddes C A, Brown D G, Fagre D B. 2005. Topography and vegetation as predictors of snow water equivalent across the alpine treeline ecotone at Lee Ridge, Glacier National Park, Montana, U.S.A. Arctic, Antarctic and Alpine Research 37. 197-205.

[5] Watson A, Davison R W, Pottie J. 2002. Snow patches lasting until winter in north-east Scotland in 1971-2000. Weather 57. 374-385.

[6] Tappeiner U, Tappeiner G, Aschenwald J, Tasser E, Ostendorf B. 2001. GIS-based modelling of spatial pattern of snow cover duration in an alpine area. Ecological Modelling 138. 265-275.

[7] Chang K-T, Li Z. 2000. Modelling snow accumulation with a geographic information system. International Journal of Geographical Information Science 14. 693-707.

[8] Purves R S, Mackaness W A, Sugden D E. 1999. An approach to modelling the impact of snow drift on glaciation in the Cairngorm Mountains, Scotland. Journal of Quaternary Science 14. 313-321.

[9] Allen T R. 1998. Topographic context of glaciers and perennial snowfields, Glacier National Park, Montana. Geomorphology 21. 207-216.

[10] Purves R S, Barton J S, Mackeness W A, Sugden D E. 1998. The development of a rule-based spatial model of wind transport and deposition of snow. Annals of Glaciology 26. 197-202.

[11] Walsh S J, Butler D R, Malanson G P. 1998. An overview of scale, pattern, process relationships in geomorphology: a remote sensing and GIS perspective. Geomorphology 21. 183-205.

[12] Kumar L, Skidmore A K, Knowles E. 1997. Modelling topographic variation in solar radiation in a GIS environment. International Journal of Geographical Information Science 11. 475-497.

[13] Watson A, Davison R W, French D. D. 1994. Summer snow patches and climate in northeast Scotland. U.K. Arctic and Alpine Research 26. 141-151.

[14] Moore I D, Grayson R B, Ladson A R. 1991. Digital terrain modelling: a review of hydrological, geomorphological and biological applications. Hydrological Processes 5. 3-30.

[15] Dozier J. 1989. Spectral signature of alpine snow cover from the Landsat Thematic Mapper. Remote sensing of Environment 28. 9 – 22.

[16] Lockwood J G. 1982. Snow and ice balance in Britain at the present time, and during the last glacial maximum and late glacial periods. Journal of Climatology 2. 209-231.

[17] Manley G. 1975. Fluctuations of snowfall and persistence of snow cover in marginal-oceanic climates. Proceedings of the WMO/IAMAP Symposium on Long-term Climatic Fluctuations. WMO No. 421. 183-188

[18] ArcGIS Solar radiation analysis sample applications: http://webhelp.esri.com/arcgiSDEsktop/9.3/index.cfm?TopicName=Solar_radiation_analysis_sample_applications

[19] ArcGIS Solar radiation analysis references: http://webhelp.esri.com/arcgiSDEsktop/9.3/index.cfm?TopicName=Solar_radiation_analysis_references