Introduction
One of the first scientific reports on slushflows was given by Reference Washburn and Goldthwait.Washburn and Goldthwait (1958). Since then, only a relatively limited number of studies have been published. During the last years, slushflow research has boon stimulated by direct observations of large events by scientists (Reference Barsch, Gucle, Mäusbacher, Schukraft, Schulte and Strauch.Barsch and others, 1993, Reference Barsch, Gude, Mäusbachcr., Schukraft and Schulte.1994; Reference SchererScherer, 1994; Reference Scherer and Parlow.Scherer and Parlow, 1994; Reference Gude and Scherer.Gude and Scherer, 1995). A short review of slushflow research has been given by Reference Gude and Scherer.Gude and Scherer (1995) and is not repeated here. Moreover, one of the most important questtons arising from the catastrophic magnitudes reached by some slush torrents, which are the high-magnitude/low-frequency variant of slushflows (cf. Reference Barsch, Gucle, Mäusbacher, Schukraft, Schulte and Strauch.Barsch and others, 1993; Reference Gude and Scherer.Gude and Scherer, 1998) is their predictability. The constructton of a prognostic slushflow model requires a solid physical basis for the initiation mechanism. It is important to mention that slushftows are frequently released by intense or long-lasting rainfalls as Reference HestnesHestnes (1985) or Reference Hestnes, Bakkehøi, Sandersen and Andresen.Hestnes and others (1994) have shown. Rainfall-triggered processes are not in the scope of this study.
One of the most important results of the Kärkevagge field campaign in the spring of 1995 (Reference Gude and Scherer.Gude and Scherer. 1995) was proof of the hypothesis that slushflows occurring during periods of intense snowmelt are primarily released due to the hydraulic gradient evolving from an increasingly inclined meltwater table within the snow cover (cf. Reference Gude and Scherer.Gude and Scherer, 1998). A video taken of the release of a slush torrent on 3 June 1995 in the Kärkerieppe cirque clearly demonstrated the dominant role of meltwater accumulatton in slushflow initiation. Figure 1 shows a photograph of the frontal wave of this slush torrent, which was followed by successive slush waves, typical features of slush torrents. The video and photographs of that slush torrent stress the extremely unsteady flood-wave character of the process. Both documents show that high amounts of water are prerequisite to enable the process.
Fig. 1. Slush torrent of 3 June 1995 released in the Kärkerieppe cirque, Kärkevagge, northern Sweden.
In this context, some general aspects of atmospheric and hydrological boundary conditions for slushflow initiation due to snowmelt are discussed below.
Study Sites
Two study sites were selected for the investigations. A small drainage basin (Kvikkåa), of 5 km2 size, was intensively studied during three field campaigns to Liefdefjorden, northwestern Spitsbergen (almost 80°N) in the spring and summer months of 1990–1992 (SPE’90–92). In each year, slushflows occurred between 5 and 14 June, despite different meteorological conditions. On 11 June 1992, several slush torrents could be observed in the Liefdefjord regton. A detailed descriptton of this study site and the events of 1992 have been given by Reference Barsch, Gucle, Mäusbacher, Schukraft, Schulte and Strauch.Barsch and others (1993), Reference SchererScherer (1994) and Reference Scherer and Parlow.Scherer and Parlow (1994).
The second study area is located in the northern Swedish mountains at almost 70°N. The Kärkevagge valley bottom has an elevatton of 700 m a.s.l., while the release zone of the slush torrent observed on 3 June 1995, is situated in a small cirque (Kärkerieppe) at almost 1000 m a.s.l. The Kärkevagge area has been well-investigated (e.g. Reference Rapp.Rapp, 1960, Reference Rapp1995; Reference NybergNyberg, 1985, Reference Nyberg.1989; Reference SchlyterSehlyter and others, 1993); the Kärkerieppe study site and instrument set-up of the field campaign in the spring of 1995 has been documented by Reference Gude and Scherer.Gude and Scherer (1995).
Methods
Field measurements of meteorological and snow-hydrological variables were analysed to determine meltwater production rates and snowpack-meltwater flow velocities. The latter aspect has been discussed in detail by Reference Gude and Scherer.Gude and Scherer (1998). Significant meltwater production occurs under isothermal snow conditions, i.e. when the total snow-pack has reached 0°C After this time, the total energy input into the snow cover is available for snowmelt. Storage heat fluxes into or from the soil or rock layer can be neglected. Radiative fluxes were measured with pyranomcters (Kipp & Zonen CMll) and pyrradiometers (Schenk 8111) at several sites. This configuration enabled determination of all components of the radiation budget, including net radiation with an accuracy of ± 10 W m−2. Turbulent-heat fluxes were measured directly using eddy-covariance systems (Gill ultrasonic anemometers, combined with fast thermocouples and hygrometers) operating at frequencies up to 21 Hz. Turbulent-heat fluxes were also determined from vertical profiles of air temperature, humidity and wind velocity. The accuracy of these measurements was ±15 W m−2.
Results
Meteorology
The occurrence of small slushflows in the Kvikkåa basin on 14 June 1991 showed that snowmelt solely due to energy input by net radiation can be sufficient to initiate minor events. Over a period of 72 hours prtor to those slushflows, the average flux density was 65 W m −2 (Reference SchererScherer, 1994), corresponding to 50 mm of meltwater production.
Prior to the slush torrents observed in 1992 in northwestern Spitsbergen and 1995 in northern Sweden, meltwater production rates had been much higher, since both net radiation and sensible-heat flux contributed to the energy available for snowmelt. Table 1 summarizes the energy-balance conditions measured at the main meteorological station of the three field campaigns at Liefdefjorden. The station is located 15 m a.s.l. near the outlet of the Kvikkåa basin. The meltwater-production rate was more than three times higher during the period prtor to slushflow release compared to 1991. In particular, high values of sensible-heat flux due to air temperatures up to 13.9°C and wind velocities of more than 10 m s−1 were responsible for 80 mm of meltwater production within 34 hours. Latent-heat flux did not contribute significantly to snowmelt.
Table 1. Energy-balance components during the 34 hours prior to the slush torrent at 16.00 CESTon 11June 1992 in the Kvikkåa drainage basin. SWE indicates snow water equivalent
The measurements during the Kärkevagge field campaign in the spring of 1995 showed that periods of intense snowmelt are partly accompanied by strong decoupling effects in the boundary layer. Figures 2 and 3 present air temperatures and wind velocities at two stations in the Kärkevagge study site. The first station (valley station, 710 m a.s.l.) was located on a flat area at the main valley-bottom, whereas the second one (cirque station, 980 m a.s.l.) was in the centre of the Kärkerieppe cirque. Prior to the slush-torrent release on the evening of 3 June 1995, this study site was almost completely covered by snow.
Fig. 2. Hourly averages of air temperatures at the meteorological stations during a period of intense snowmelt in the Kärkevagge, 1995.
Fig. 3. Hourly averages of wind velocities at the meteorological stations during a period of intense snowmelt in the Kärkevagge, 1995.
Despite the short distance between both stations (1 km horizontal, 270 m vertical), air temperatures differed more than 5 K with higher values in the cirque area, where the slush torrent was released.
In contrast to the events on 11 June 1992, in the Liefdefjord regton, the Kärkevagge slush torrent was not released during extreme weather conditions. Nevertheless, warm-air advection took place, which strongly affected the higher-cirque area, where the slush torrent was released. Figures 2 and 3 demonstrate that critical meteorological conditions can lead to a high spatial variability of the energy balance even in small drainage basins. A spatially distributed treatment of the atmospheric variables is probably required for a prognostic model that could be applied for risk-assessment purposes of forecsting tasks.
Both slush torrents were released while snowmelt was driven by net radiation in combination with the sensible-heat flux. Sensible-heat flux is important, since net radiation is limited by astronomical factors and by snow albedo, which was higher than 65% in both cases.
High meltwater-production rates leading to 50–80 mm meltwater production within a few days was shown to be essential for the initiation of slush torrents. However, even such high melt rates are not able to generate water-saturated layers up to several meters thick as observed in Kärkevagge solely from local snowmelt. Meltwater flow through the snowpack is the link between the meteorological conditions and slushflow release.
Snow Hydrology
Snow—hydrological conditions are not only dependent on meteorological conditions but also on terrain features in the initiation area and the adjacent slopes. The basic correlation between topography and snow processes is stressed by a comparative analysis of the investigated processes in northwestern Spitsbergen and northern Sweden. Despite the fact that the Kvikkåa basin is more than 1000 km closer to the North Pole, both regions show many similarities:
Snowmelt is concentrated in one pronounced period starting in late-May and lasting approximately 1 month. Onset of snowmelt was delayed 1 or 2 weeks in the Kvikkåa basin compared to the Kiirkcvagge basin during the years of the investigations.
During the main snowmelt period, solar irradiance may occur 24 hours per day.
Slushflow release zones are situated in gently inclined sections of valleys, which are part of the fluvial system during die summer.
In close proximity to the release zones, the up-slope area includes sleeper slopes of significant extent.
There are several hydrological implications resulting Irom these similarities.
A snow cover of at least 40–50 cm depth must be present, when solar irradiance is close to its annual maximum. Despite the high solar albedo of snow, higher values of solar irradiance will successively lead to an increasing energy input by net radiation into the snow cover. During the same period, the probability of warm air advection is increased. Both factors are needed for a superposition of high values of net radiation and sensible-heat flux.
Interrupiton of meltwater production or refreezing of meltwater during the night should not occur significantly within the snowpack. These processes have two stabilizing effects. On the one hand, they cause a certain delay of the meltwater production during day-time. Also, the distance that meltwater can pass by flowing through the snowpack is shortened.
Meltwater flow concentrates in the valley bottoms. On steeper slopes, drainage conditions are usually sufficient to prevent meltwater accumulation due to the higher hydraulic gradients compared to gently inclined sections.
Steep and extended slopes adjacent to the slushflow release zone are favorable topographic situations, particularly for slush torrents, since the typical duration of an intense snowmelt period is only a few days. Meltwater produced during such periods must be able to flow through the snow to reach these areas, where convergence takes place. Considering meltwater-flow velocities of less than 1 m h−1 in the beginning and 20–40 m h−1 as an upper limit of snowpack, meltwater-flow velocity under saturated conditions at slopes (cf. Reference Gude and Scherer.Gude and Scherer, 1998), it is obvious that the maximum distance between meltwater production and accumulation is limited. Assuming a high upper limit of the average velocity of 20 m h−1 and a duration of the snowmelt event of 50 h, meltwater could pass a distance of 1 km. This demonstrates that a maximum distance of about 1 km could be used for determining the catchment area of a slushflow release zone.
Slushflows are frequently released in terrain situations with confluent meltwater pathways (cf. Reference NybergNyberg, 1985). It is possible to identify these critical areas by a digital morphometric analysis. However, confluence is an insufficient condition for slushflow release. Due to the distributed nature of meltwater production and drainage through the snow matrix, meltwater flow has to converge to generate an inclined water lable. High rates of meltwater convergence are only possible as long as the bulk hydraulic conductivity of the snow cover is small, at least in the release zone. Otherwise, the outflux of meltwater at a given vertical snow column would not be over compensated by the influx. Areas of convergent meltwater flow can mainly be observed in valley bottoms of gentle slope angles prior to the onset of channelled runoff.
Significant infiltration losses would prevent meltwater accumulation. In polar and sub-polar drainage basins, there are several factors reducing infiltration of meltwater in the sub-layer.
Frozen ground usually acts as an effective impermeable layer, either by a very low hydraulic conductivity or by causing basal refreezing of meltwater. If permafrost conditions prevail, the sub-layer is frozen as long as snow is present. In polar and sub-polar regions, the sub-layer is usually frozen during snowmelt, even when permafrost is not present.
Many drainage basins are characterized by weakly developed or an absence of soils. Impermeable rock layers can substitute frozen ground as inhibitor of infiltration losses.
Conclusions
Slushflows can be released due to energy input in the snow cover by net radiation and sensible heat within the ordinary range of a high-latitudinal snowmelt period. Sensible heat flux is particularly important for slush-torrent initiation, since it is not limited by external factors as is the case for net radiation. A superposition of high net radiation and sensible-heat flux is most favorable for high-magnitude events.
With respect to slushflow initiation due to snowmelt, infiltration losses can be disregarded at high latitudes, even when permafrost is not present. This explains why slushflows are not restricted to permafrost areas.
Crucial for slushflow initiation, due to snowmelt, is the timing of energy input and meltwater flow through the snowpack. Gently sloping sections with confluent meltwater paths in valley bottoms and steeper slopes in the adjacent up-slope area are well-disposed for slush-torrent initiation. The distance between meltwater production and accumulation is limited by the duration of the intense snowmelt period and by the velocity of the meltwater flow through the snowpack.
Acknowledgements
We wish to express our thanks to all those who have helped us to obtain these results, particularly to D. Barsch, W.-D. Blümel, G. Bing, T. Löschmann, R. Mäusbachcr, A. Rapp, N Ritter, A. Schulte, G. Schukraft and F Siegrist. The Abisko Scientific Research Station has supported our research in the Kärkevagge logistically. Projects contributing to these results were financially supported by the Swiss National Science Foundation, the German Science Foundation and the Swiss Academy of Science.
