University of Fribourg American University of Armenia National Academy of Science, Armenia Cage Holding SA FREEZWATER Final report Hydrological potential of snow and artificial snow preservation on Mount Aragats, Armenia Alexander Nestler Matthias Huss Fribourg, October 2, 2013
On Armenia s highest peak, Mount Aragats, large, wind-driven snow accumulations are covered with Geotextiles in order to delay snow melt and to provide additional water in the drier period of the year. Between 2011 and 2013 the University of Fribourg has performed extensive glaciological investigations related to the project FREEZWATER. This final report provides proofs for the favorable effect Geotextiles have upon the redistribution of seasonal runoff from snow-covered areas. Uncovered snow has almost disappeared while textile-covered areas are shown to still provide significant runoff. However, our studies also emphasize the necessity for Geotextile coverage on a larger scale in order to achieve a significant effect. Calculations are performed with an energy-balance model and with a distributed temperature-index model. Both models coincide in their statement about the efficiency of Geotextiles. Nevertheless, extrapolations to higher altitudes yield diverging results, attributing a better performance to Geotextiles using the temperature index model. In spite of these uncertainties, this study provides insights into the processes of wind-driven snow redistribution and the nature of snow melt in high mountain regions and helps to estimate the effect of snow preserving methods, which are performed in the context of the Swiss- Armenian FREEZWATER Project. This report is based on the Master thesis by Alexander Nestler that contains more detailed information and further explanations. Introduction Mount Aragats Armenia s highest mountain is subjected to large quantities of snow every winter. These snow accumulations are redistributed by strong winds and are deposited on the lee side of sharp edges in topography as snow cornices. These cornices persist when most of the general snow cover has disappeared in early summer. They still provide melt water on a regional scale when temperatures rise and amounts of precipitation drop. Nevertheless, by the arrival of the driest period, in mid-august, the snow cornices have already disappeared. All streams descending from Mount Aragatz dry up as they are mostly fed by melt water. Since the underground mainly consists of large volcanic boulders with a high permeability for water, these streams dry up in a matter of days leaving the Aragats region with a severe water deficiency. With the expected future changes in the global climate system the hydrological stress is even expected to increase over the next decades. Against this background, members of the Swiss-Armenian joint venture FREEZWATER cover those wind-driven snow accumulations in an attempt to slow their melting sufficiently to preserve them long enough to provide melt water when it is most needed. This study scientifically accompanies an experimental coverage of one particular snow cornice with Geotextiles. In three extensive field visits to Mount Aragatz between 2011 and 2013 glaciological measurements were performed in order to estimate the efficiency of the application of Geotextiles. The effect Geotextiles have upon the melting of snow fields was already intensively addressed in several scientific publications (see e.g. OLEFS and FISCHER 2008 and OLEFS and LEHNING 2010). Our study combines the field observations close to Lake Vishab with the numerical glacio-hydrological model GERM (HUSS et al. 2008) in order to provide specific estimates for the efficiency of snow preservation methods and their impact on runoff from Mount Aragats. With the help of GERM both the spatial evolution and the generated runoff was calculated based on an energy-balance approach (OERLEMANS 2001), as well as a distributed temperature-index approach (HOCK 1999). Both approaches were compared and thus
provided not only a validation of the results but also a defined uncertainty range of possible simulation outcomes. Furthermore, LANDSAT satellite imagery was used to assess the spatial distribution of snow over the entire Mount Aragats. For an application in mountainous areas where no adequate satellite imagery is available, the distribution of snow was assessed using two different models. These deterministic models involved statistical as well as process-based approaches to determine snow distribution and quantity, and additionally provided a rough estimation of the total snow volume on Mount Aragats at a given point in time. In brief, this study will on the one hand be a help to guide and evaluate the activities performed in the context of the FREEZWATER Project. It will on the other hand also provide insights into processes governing the formation of wind-blown snow accumulations and the nature of snow melt in high mountain regions. Literature review about melt reduction and frozen water storage Debris coverage Layers of debris are known to affect the energy balance of underlying snow surfaces (KAYASTAH et al. 2000). Thin layers of debris will increase snow melt as a result of an increased surface albedo, which contributes to a heating of the surface (KAYASTAH et al. 2000, TAKEUCHI et al. 2000). With increasing layer thickness, however, less energy is transported through the debris layer down to the snow surface. At a critical thickness of around 2-5 cm the isolation effect of a debris layer outweighs the heating effect caused by a higher albedo (KAYASTAH et al. 2000). The critical depth depends on the type of debris. Beyond the critical thickness, a debris layer slows down the melt process. KAYASTAH et al. 2000 even claim a complete stop of melt processes with layer thicknesses of 1 m. Artificial snow surface coverage Already in the 1960 and 70s different techniques were tested to slow down the melt of snow and ice. Both saw dust and aqueous foams were applied on snow surfaces and reduced melt rates significantly (HERRMANN and STEHLE 1966). Although saw dust proved to have less insulation capacities, it was found to have a cooling effect on the surface through the evaporation of absorbed water. Nevertheless, both saw dust and aqueous foams are prone to be blown out by wind. In a recent study by SKOGSBERG and LUNDBERG 2005 larger woodchips were tested as a protection of snow accumulations. A layer thickness of 7.5 cm was reported to reduce snow melt by 85 %. The most efficient way of slowing snow melt was already presented in 2008 by OLEFS and FISCHER. After testing several different materials and techniques, a protection with Geotextiles turned out to have the most favorable impact on the melt regime. In the course of one melt season, these Geotextiles were found to reduce melt rates by up to 60 %. A doubling of layers reduced melt rates by further 10 %. However, adding additional layers did not further reduce melt rates. Artificial frozen water storage Water cannot be stored only by protecting natural snow or ice but also by artificially creating accumulations of snow and ice. In Ladakh, India, a series of low walls dam up water from a small river to form shallow accumulations of ice in winter time (NORPHEL 2009). Different approaches involve a water
spray from an elevated point freezing into a block of ice during winter time. Fig. 1 shows an experiment on Mount Aragats, Armenia. These objects are termed artificial glaciers, although they do not live long enough (till the end of July) to be called actual glaciers. Fig. 1 Artificial glacier created on Mount Aragats in 2010 (photo: A. Hambarian) Study site The investigated snow fields are situated on the southern slopes of Mount Aragats, Armenia, at an elevation of 3200 m a.s.l. Several wind-blown snow cornices along edges in the topography (Fig. 2) were investigated at four sites in the vicinity of the Cosmic Ray Station Aragats (Fig. 3). On one snow cornice close to Vishab Lake experiments with Geotextiles provided by Landolt AG, Näfels, Switzerland, were performed in 2012 and 2013 by the members of the FREEZWATER project under the guidance of Prof. R. Ambartsumian. Fig. 2 Snow cornice along the Shampoor ridge on 04 July 2011 (photo: A. Nestler)
Fig. 3 Overview of the area of investigation. Snow cornices classified from LANDSAT image on 11 June 2011 Data In three extensive field visits in 2011, 2012 and 2013, field data was obtained that included GPS-perimeter mappings, snow depth measurements, snow density measurements and ablation measurements (measurements of the speed of melt). Directly within the area of investigation weather data was obtained from the Cosmic Ray Station. Data series reach as far back as June 2011 and include: air temperature, wind speed, wind direction, precipitation rate and solar radiation at hourly resolution. For all topography-related analysis, an ASTER GDEM (Global Digital Elevation Model) was used. Furthermore, snow distribution was derived from a LANDSAT satellite image capturing Mount Aragats on 11 June 2006. Methods In order to simulate the development of snow accumulations over time the glaciohydrological model GERM (Glacier Evolution Runoff Model) developed by HUSS et al. 2008b was used. The calculation of mass balances within GERM is based on a simplified energy balance approach (OERLEMANS 2001), as well as on the a distributed temperature-index model (HOCK 1999). The energy balance model tries to capture the physical processes involved when it comes to snow melt. In the energy balance approach, GERM calculates cumulative mass balances (B cum ) at the snow surface in meters water equivalent (m w.e.) for each grid cell (1 m
resolution) and for every time step (t+1) of the respective day individually as follows (see e.g. MACHGUTH et al. 2006): (1) In this formula time in a discrete interval of one hour is represented by the variable t. L is the latent heat energy available for melting. E is the energy flux at the surface averaged over one day. Accumulation is accounted for as solid precipitation (P SOLID ). The temperature index approach is based on the assumption that positive temperatures are closely related to snow melt (FINSTERWALDER and SCHUNK1887). As suggested by HOCK 1999, melt rates M in mm d -1 are hence calculated as: (2) where MF is a factor describing melt in mm d -1 C -1. As indicated by its unit this factor is calculated on a daily basis. In order to scale it for hourly time steps it is divided by the number of time steps per day (n = 24). r fsnow is a radiation coefficient for fresh snow and r osnow refers to old snow. I pot describes the potential clear-sky direct irradiance from the sun (W m -2 ), which is calculated in a subroutine of GERM. T is the air temperature in C. In this approach, melt is assumed to be zero for temperatures below 0 C. Both models were calibrated adapting parameters in a way that calculated melt rates matched measured melt rates. Results A comparison of melt rates observed on Geotextile covered areas and on snow covered areas demonstrates how drastically Geotextiles reduce the melting of snow on Mount Aragats (Fig. 4). Fig. 4 Cumulative snow melt in June 2013. Solid lines represent measured melt below Geotextiles, dashed lines melt in uncovered parts of the snow field.
A single Geotextile layer was thus found to reduce melt rates by 57 %. A double layer even leads to a melt reduction of 75 %. Measured Snow densities were between 590 kg m -3 (2012) and 639 kg m -3 (2013). Fig. 4 shows perimeter mappings of the same snow cornice (Level 01) performed in 2011, 2012 and 2013. These mappings reveal annually recurring patterns, which indicate that snow distribution on Mount Aragats follows the same physical processes in every winter and can therefore be estimated accurately using deterministic (or stochastic) models. Using GERM the survival of snow accumulations was modeled with regard to the location of the snow fields (altitude above sea level, exposure to the sun) and the Geotextile coverage (Fig.6 and 7). For an elevation of 3200 m a.s.l. both the energy-balance model and the temperature-index model calculate similar longevities for snow masses. An uncovered snow field with an initial thickness of 7 m, for instance, is likely to have disappeared by 6 to 9 August 2012 after between 47 and 50 days after of melting. The same snow accumulation at the same elevation covered with Geotextile is however expected to completely melt only between 16 and 18 October 2012. It thus survives more than two months longer. If the snow field is located at higher altitudes (i.e. above Lake Vishab) some differences between the two melt models are evident. Uncovered snow masses of 7 m depth at 3950 m a.s.l. are calculated to last till 17 August 2012 as simulated with the energy-balance approach. In contrast, Fig. 5 Comparison of perimeters of Level 01 mapped in 2011, 2012 and 2013 the temperature-index approach suggests a survival till 29 August 2012. The spread of results is significantly amplified when melt is simulated beneath Geotextile coverage (Figs 6 and 7). According to the temperature-index model, melt at 3450 m a.s.l. decreases rapidly towards the end of summer. Melt processes more or less stop on 18 October 2012. Thus, the Geotextile provides sufficient protection to keep a snow mass of 6 m depth from completely melting within the melt season. The energy-balance model suggests that even snow masses with snow depths of up to 7 m are likely to disappear completely within the summer season. Only at 3700 m a.s.l. does the energy-balance model suggest a complete survival of snow masses with a depth of 6m. Our experiments indicate that some uncertainties in the modeling of snow melt for conditions with or without Geotextile protection are present, but that using artificial protection, snow fields with a certain initial depth (about 6-7m) are likely to survive the hot summer season on Mount Aragats. This is consistent with the in-situ experiments performed within the FREEZWATER project in 2012 and 2013.
Fig. 6 Survival of snow masses with regard to initial snow depth on various levels of altitude integrating effects of textile coverage (c: covered; uc:uncovered) modeled with the temperatureindex model Fig. 7 Survival of snow masses with regard to initial snow depth on various levels of altitude integrating effects of textile coverage (c: covered; uc:uncovered) modeled with the energy-balance model.
Runoff generated by snow melt is a further feature calculated by GERM. A comparison of simulations with and without coverage reveals a significant redistribution of runoff from June/July into August and September under the assumption that the whole snow field is covered (Fig. 8). Fig. 8 Daily runoff at 3200 m a.s.l. simulated for covered snow surfaces (c) and uncovered snow surfaces (uc) with the temperature-index model (TIM) between 17 June 2012 and 16 December 2012. Conspicuous peaks in the runoff distribution are linked to precipitation events. Till mid-july runoff from uncovered areas is 2 to 3 times greater than from covered areas. Beyond that point, runoff from covered areas is higher than from uncovered areas. The resulting runoff from covered areas towards the end of August is 5 times greater than it would have been without Geotextile coverage. On 27 August 2012, for example, GERM calculates a runoff for uncovered areas of 20 m 3 day -1. On the same day the calculated runoff from areas covered with Geotextiles amounts to 108 m 3 day -1 (Fig. 8). To achieve such a benefit, the area covered with Geotextiles would need to be increased by a factor of 40 compared to the experiments performed in 2012 and 2013. Discussion Recurring patterns of spatial snow distribution observed in the perimeter mappings indicate annually repeating processes. Minor differences such as tongues reaching down from the cornices might be caused by random processes such as ruptures in the snow cliff triggering avalanches (Fig. 5). One major difference between the energy-balance model and the temperature-index model applied for calculating snow melt consists in their performance below 0 C. The energybalance model may calculate melt for temperatures below the freezing point, which is not possible with the temperature-index model. This difference is most likely responsible for the spread in the results between the models for higher elevations where temperatures remain more often below 0 C (Figs 6 and 7). Simulations with Geotextile coverage show a significant redistribution of runoff compared to bare snow surfaces. Fig.9 illustrates the magnitude of this effect as the difference between runoff from covered and uncovered areas. Clearly, three phases are discernible that characterize the effect Geotextiles have upon runoff. In the first phase (I-a) melt rates in uncovered exceed those in covered areas. As a consequence runoff from uncovered areas is calculated to be up to 800 m 3 day -1 higher compared to covered areas. The higher melt rates lead to a much faster decrease of snow volume in uncovered regions. This results in a rapid decrease in runoff from uncovered snow accumulations as compared to covered ones until at one point runoff from covered areas exceeds runoff from uncovered areas.
During phase I-b the difference between runoffs in favor of runoff from covered snow accumulations steadily increases. The surplus of runoff from covered areas gets bigger until the entire uncovered snow accumulation has melted away around 03 August. This is the point where the snow covered with Geotextiles produces the maximum benefit in terms of runoff. According to the model up to 200 m 3 day -1 of additional runoff from covered areas during this period can be expected. After the complete disappearance of uncovered snow masses the difference between the two runoff configurations diminishes as runoff from covered areas decreases with decreasing snow volumes and falling air temperatures (phase II). This process continues till all remaining snow masses under Geotextile coverage have disappeared (phase III) (Fig. 9). complete meltdown uncovered area complete meltdown covered area Ia Ib II III Fig. 9 Difference between runoff from covered and uncovered snow accumulations on 3200 m a.s.l. simulated with the distributed temperature-index model Fig. 9 illustrates the significant redistribution of runoff as an effect of covering a snow field of a typical size with Geotextiles yielding up to 200 m 3 of additional water melt on days when natural uncovered snow is almost completely gone. Nevertheless, the investigated effects concern a complete coverage of the entire test area which corresponds to about 20 000 m 2. In 2012 and 2013, only 2-3 % of this area: 450-600 m 2 were covered with Geotextiles. The effects of coverage on such a small scale are correspondingly lower, providing 6-7 m 3 day -1 during the redistribution peak. Hence, for one square meter of Geotextile, the maximum difference in runoff redistribution is 10 liters day -1. Snow ablation measurements demonstrated the strong melt-reducing effect of Geotextiles. By contrasting daily melt rates of covered and uncovered areas and their integration over a defined area it is possible to determine the water-saving capacity of Geotextiles. With a melt reduction of 57 % that was found during the measurements, one square meter of Geotextile is capable of saving 0.04 m 3 of water per day. For one piece of Geotextile with an area of 250 m 2 this means that 9 m 3 less water melts away every day and can thus be stored. The maximum difference between covered and uncovered snow volumes can be found at the beginning of August when the last remainders of uncovered snow disappear. The water that is saved until this day for one m 2 of Geotextile is 1.94 m 3, which corresponds to 484.5 m 3 for on slab of Geotextile. Regardless of the area covered, a protected snow surface will always persist considerably longer than it would without any coverage. This circumstance gave rise to the idea of preserving a strip of snow to create a piste for skiing in autumn when most tourists visit the Aragats area. This strip of snow is supposed to be as long as possible reaching from the South Saddle all the way down to the Lake Vishab as shown in the terrain model in Fig. 10. With an assumed width of 20 m and a length of 3.7 km this piste would cover an area of 75 140 m 2. An area of that size would require about 300 individual pieces of Geotextile. From a financial point of view it is doubtable whether the revenues of the skiing business would make up for the enormous costs of the Geotextiles that would have to be imported from
Switzerland. Furthermore, at the moment, there are no adequate machines available on Mount Aragats to place those textiles in an efficient way. Placing them manually would require several hundred workers to be paid for one week or more. However, there is an alternative way to covering a continuous strip of snow to create a skiing piste: By the end of June, snow fields still cover all the way up to the southern peak of Mount Aragats in large, partly isolated patches. These patches are separated by narrow corridors of rock. If Geotextile coverage were applied early in the season on top of the snow where later the corridors begin to form, the rock corridors can be prevented from opening. This N Fig. 11.3 Hypothetical skiing piste, symbolized by a white strip, reaches from the south saddle at 3700 m a.s.l. down to Lake Vishab at 3200 m a.s.l. 3D visualizations based on ASTER GDEM. elevations are exaggerated would on the one hand not only consume much less Geotextiles, but on the other hand would create a much larger coherent area for skiing that would be operational from the opening of the road in mid-may until the end of July or even longer. Instead of actively protecting snow accumulation it may be a lot more time and cost-efficient to imitate the processes contributing to the formation of snow cornices. The pushing of snow masses over the edges of canyons would have the same effect as wind-driven relocation of snow. In contrast to placing Geotextiles, this method would be feasible for a small number of workers in a comparably short time under the condition of having access to machines such as snow groomers. Snow could be compacted with this method as well increasing its longevity with a given depth. Another possibility to enhance snow accumulation in the lee side of edges would be raising the edges themselves. This could be achieved by constructing walls along the edges using volcanic boulders that are almost omnipresent within the area. However, the efficiency of this method on Mount Aragats would need to be first investigated. In terms of efficient snow preservation a combination of several activities would certainly optimize the outcome. Conclusion During three field visits to Mount Aragats, Armenia, between 2011 and 2013 glaciological data was collected to investigate snow preserving activities performed in the context of the Swiss-Armenian FREEZWATER Project. Snow preserving activities involve the coverage of natural wind-driven snow accumulations with Geotextiles. Measurements of melt rates showed that the Geotextile coverage reduces snow melt by about 60% (and even more in some cases). The method is therefore considered as highly efficient. By means of numerical modeling, the efficiency of Geotextiles in terms of snow preservation such as the spatio-temporal evolution of snow fields could be calculated. Results were extrapolated to other years and different altitudes on the mountain. Results were both calculated with a process-based energy-balance model and a distributed temperature-index model. Both models show in accordance with observations that several meters of snow disappear around the end of July / beginning of August without Geotextile coverage, even at higher elevations. For Geotextile-covered snow accumulations however which feature a depth of 7 m in mid-july survival till the beginning of October is probable
according to both models. An extrapolation of results for covered snow fields to higher altitudes yields diverging results between the models. While according to the temperatureindex model 3.5 m of snow suffice to be preserved throughout the whole year at 3950 m a.s.l., the energy-balance model calculates a minimum of 5.5 m of snow to be preserved under the same conditions. A strong hydrological impact of Geotextile coverage is revealed by comparing modeled runoff from covered and uncovered snow fields. By the end of August, when air temperatures are highest, Geotextile-covered areas yield, according to the model, five times more water than uncovered snow accumulations. Significant effects can, however, only be observed for snow coverage at a larger scale. An area as covered in 2012 (600 m 2 ) would yield 7 m 3 of additional water per day during August. With the guidance of this study, the water situation in the region of Mount Aragatz will hopefully one day be improved with melt water originating from protected snow accumulations. For the University of Fribourg, Alexander Nestler, M.sc. Dr. Matthias Huss Selected references BOJANOWSKI, A. 2010. Kampf gegen Klimawandel-Folgen. Forscher züchtet Gletscher im Schwarzwald. Spiegel Online. CUFEY, K., PATERSON, W. 2010.The Physics of Glaciers. Elsevier, Inc., 4th edition FARINOTTI, D., USSELMANN, S., HUSS, M., BAUDER, A., FUNK, M. 2012. Runoff evolution in the Swiss Alps: projections for selected high-alpine catchments based on ENSEMBLES scenarios. Hydrol. Process. 26. 1909-1924. FINSTERWALDER, S., SCHUNK, H. 1887. Der Suldenferner. Zeitschrift des Deutsch- Österreichischen Alpenvereins 18. 70-89. HERRMANN, M., STEHLE, N. 1967. Protective Coverings for Ice and Snow. Physics of Snow and Ice: proceedings, 1(2): 797-806. HOCK, R. 1999. A distributed temperature-index ice- and snowmelt model including potential direct solar radiation. J. Glaciol., 45(149), 101 111. HUSS M, BAUDER A, FUNK M, HOCK R. 2008a. Determination of the seasonal mass balance of four Alpine glaciers since 1865. Journal of Geophysical Research 113: F01015. HUSS, M. (2011) Present and future contribution of glacier storage change to runoff from macroscale drainage basins in Europe. Water Resources Research 47, doi:10.1029/2010wr010299. HUSS, M., FARINOTTI, D., BAUDER, A., FUNK, M. 2008b. Modeling runoff from highly glacierized alpine drainage basins in a changing climate. Hydrol. Process. 22. 3888-3902 KAYASTHA, R., 2000. Practical prediction of ice melting beneath various thickness of debris cover on Khumbu Glacier, Nepal, using a positive degree-day factor. Debris-Covered Glaciers. JAHS Publications 264. 71-81. KLOK, E., OERLEMANS, J. 2002. Model study of the spatial distribution of the energy an mass balance ofmorteratschgletscher, Switzerland. Journal of Glaciology 48(163). 505-518 MACHGUTH, H., PAUL, F., HOELZLE, M., HAEBERLI, W. 2006. Distributed glacier massbalancemodelling as an important component of modern multi-level glacier monitoring. Annals of Glaciology 43. 335-343 NORPHEL, C. 2009. Artificial glaciers: a high altitude cold desert water conservation technique. Energy and climate change in cold regions of Asia. Proceedings of the Seminar.62-70. OERLEMANS, J. 2001. Glaciers and climate change. Lisse, etc., A.A. Balkema. OLEFS, M., FISCHER, A. (2008). Comparative study of technical measures to reduce snow and ice ablation in Alpine glacier ski resorts. Cold regions science and technology, 52(3), 371-384. SKOGSBERG, K., LUNDBERG, A., 2005. Wood chips as thermal insulation of snow. Cold Regions Science and Technology 43, 207 218.