The role of snowpack in producing floods under heavy rainfall

Transcription

1 Hydrology, Water Resources and Ecology in Headwaters (Proceedings of the HcadWater'98 Conference held at Itféran/Merano, Italy, April 1998). IAHS Publ. no. 248, QQ y The role of snowpack in producing floods under heavy rainfall PRATAP SINGH National Institute of Hydrology, Roorkee , Uttar Pradesh, India GERHARD SPITZBART, H. HUBL & H. W. WEINMEISTER Institute for Torrent and Avalanche Control, BOKU, Peter Jordan Strasse 82, A-1I90 Vienna, Austria Abstract In order to understand the hydrological response of rain-on-snow events, artificial rains with different intensity and duration were simulated on a snow plot consisting of natural snowpack, and the behaviour of the emerging outflow was studied. This study was carried out in the Austrian Alps. Measurements of meteorological parameters, soil temperature and snowpack properties were also made. It was found that the presence of ice layers increased the liquid water holding capacity of the snowpack to more than two times that of an homogeneous snowpack. Under the saturated conditions of the snowpack, rainwater moved with a speed of about 6 m h" 1 suggesting that the speed of rainwater may be several times higher than the normal snowmelt transmission through the snowpack under non-rainy conditions. Most of the input appeared as runoff under heavy rainy conditions because the snowpack was quickly saturated and conditioned to yield the maximum runoff. INTRODUCTION Rain-on-snow events are considered important for producing high streamflows, triggering avalanches, landslides and debris flows. Such events have much greater potential for generating' serious floods than short periods of radiation-induced snowmelt (Kattelmann, 1987). Most of the largest floods in British Columbia, Washington, Oregon and California have been associated with rain-on-snow situations (Kattelmann, 1987; Brunengo, 1990). In Britain more frequently flooding results from a combination of melting snow and rainfall (Archer et al., 1994). Moderate rainfall is observed in the high altitude regions of Himalayas during the active snowmelt period, but its impact on generating floods is to be studied (Singh et al, 1995; Singh & Kumar, 1997). Various studies are carried out to understand the role of rain-on-snow events in triggering avalanches (Conway et al, 1988; Conway & Raymond, 1993) and landslides (Harr, 1986; Christner & Harr, 1982; Bergman, 1987). Further, Sandersen et al. (1996) reported that the combination of rainfall and snowmelt increased the frequency of debris flows in Norway. An accurate estimation of timing, amount and rate of outflow is required to improve streamflow forecasting for reservoir operation, flood control and design of major structures. To include rain-on-snow events in the modelling of runoff, a knowledge of liquid water storage, natural and rain-induced melting, and transmission characteristics through the snowpack is needed. Most of the earlier

2 90 Pratap Singh et al. studies were carried out to understand the processes of meltwater percolation normal snowmelt runoff through the snowpack (Colbeck, 1972, 1974; Colbeck & Davidson, 1973; Dunne et al, 1976; Wankiewicz, 1976). Under heavy rainy conditions, the response of a snowpack will differ from that during normal snowmelt. The present study is carried out to understand the response of snowpack under heavy rainy conditions and quantitative experimental data obtained through field investigations are presented. To carry out this study, artificial rains with different intensity and duration were simulated over two snow plots from the natural snowpack in the Austrian Alpine region. STUDY AREA AND SETTING OF THE EXPERIMENT The study was conducted in the Glatzbach watershed located about 5 km southwest of Grossglockner, the highest peak of the Austrian Alps. The experiment was conducted on 13 May 1996 and the depth of the snowpack was 1.08 m at the experiment site. An isolated snow plot was prepared in natural field conditions at an elevation of 2640 m. The dimensions of the plot were 2.30 x 1.30 x 1.08 m maintaining the natural structure of the snowpack. To check the lateral flow, three sides of the plot were wrapped by a plastic sheet and the runoff draining side was left exposed. The upper surface (air-snow) and lower surface (ground-snow) were not disturbed. SIMULATION OF RAIN AND OBSERVATIONS OF RUNOFF AND OTHER METEOROLOGICAL PARAMETERS A rainfall simulator developed at the Institute for Torrent and Avalanche Control, Vienna, Austria, was installed to simulate rainfall over the snow plot. The rain drops fell over the snowpack from a height of about 1 m. The rainfall simulating device moved both in a forward and backward direction with a constant speed of 0.20 m s" 1 over the snow plot providing a uniform distribution of rain over the study plots. In the present study, different intensities were used for different events. Runoff from the plot was measured by employing a tipping bucket device. The surface and subsurface flow were collected in an open pipe channel immediately after emerging from the plot and drained into the tipping bucket. Continuous measurements of net radiation, air temperature, humidity, and wind velocity were made at 2 m above the snow surface. No precipitation was observed during the experiment and most of the time cloudy conditions prevailed. The stratigraphie observations of the snowpack included height, density, ice layers, grain size and type, liquid water content and temperature profile. The density of the snow was determined using sampling snow tubes of known volume and the liquid water content was measured with a dielectric moisture meter with a flat capacitive sensor. The meteorological records and changes in soil moisture during the experiments are shown in Fig. 1. The sky was overcast during the experiments and air temperature was not too high. These weather conditions provided the environment very near to the actual rainy conditions in the study area.

3 The role ofsnowpack in producing floods under heavy rainfall , 6 -, _ 60 -, "5 > 50 o i 30 'o I ' ' ' I ' Fig. 1 Net radiation, air temperature, relative humidity, wind speed and soil moisture content observed on 13 May in the Glatzbach basin, in the Austrian Alps. RESULTS AND DISCUSSION The study was carried out on a snow plot for two rain-on-snow events with different rain intensities. The average rain intensity, starting time and duration for the different events for each plot are given in Table 1. Because the objective of the study was to study the runoff response of the snowpack under heavy rainfall events, high rainfall intensities were used. The first rainfall simulation was made at 1130 h and continued for a period of 2 h with an average intensity of 1.27 mm min" 1. The snow plot was isothermal at 0 C, therefore, any heat deficiency can be considered to be Table 1 Details of the different rainfall simulations over snow plots. Rainfall event Starting time Average rainfall intensity (mm min" 1 ) Duration (min) Event 1 Event h 1430 h

4 92 Pratap Singh et al..a s Rain event 1 Rain Runoff Snowm eit E ! ' i r ~r~ i i i i i Minutes after rainfall started at 1130 hours Rain event , Rain Runoff S n ow m elt E I ' I ' I ' I ' I ' I ' I ' I ' I Minutes after rainfall started at 1430 hours Fig. 2 Simulated rainfall, computed snowmelt and observed runoff from a snow plot. negligible. Before starting the rainfall, the average liquid water content of the snowpack was about 4% by volume. The snowpack contained five well-distinguished ice layers ranging from about 2 to 8 mm in thickness. The distribution of the rainfall input and the outflow from the plot for this rainfall event is shown in Fig. 2. In spite of the isothermal state of the plots and heavy rainfall inputs, negligible runoff was observed from the plot and this runoff was from the subsurface flow only. During this event, the total depth of rainwater supplied to a unit area of the plot was 155 mm. No significant change was observed in the soil moisture content under the snowpack. This confirms that the total quantity of rainwater was absorbed and retained in the snowpack only.

5 The role ofsnowpack in producing floods under heavy rainfall 93 The second rain simulation was conducted at 1430 h i.e. 1 h after stopping the first rainfall simulation event. This event was also continued for a period of 2 h with an average rainfall intensity of 1.66 mm mm 1. The runoff from the plot was measured until the complete recession of outflow and illustrated in Fig. 2. Unlike the first event, the snow plot started to produce runoff soon after the rainfall. The water percolation rate and the response time of the water flowing under the snow to the stream channel, are principal factors determining the time of arrival t a and time of equilibrium concentration t e of liquid water in snow. These two factors are mainly governed by the structure, transient storage and water flux to the pack. The time of equilibrium concentration represents the time of peak streamflow when input is constant and equal to the time required for water to reach the stream from the furthest part of the watershed. The lag-time in producing runoff was about 10 minutes, whereas time of equilibrium concentration was about 30 minutes. It shows that water infiltrated very fast from both plots. The duration of rainfall t r was greater than t e Therefore, the outflow reached a constant value, slightly higher than the rainfall input. Runoff higher than the rainfall input can be expected because of the snowmelt contribution. Some natural snowmelt also resulted during this rainfall simulation period and a little rain-induced melt also occurred during the rainfall simulation period. The natural and rain-induced snowmelt was computed using the energy balance approach. It was noticed that abundant rainfall could not produce snowmelt was not significant. This was because of very little energy supplied to the snowpack through rainfall as compared with other components of the energy source. During the experiments, the snowpack was at 0 C. Therefore, the possibility of refreezing the water can be ignored. The total rainfall remained in the liquid form in the snowpack. This snowmelt added to the rainfall runoff and emerged as combined outflow. The recession of the hydrograph started soon after stopping the rainfall and reached a very low outflow. The major part of the recession occurred within minutes after stopping the rainfall, but a slow recession prolonged for about 1 h. This was because of the major contribution of rainfall in the outflow. Measurement of soil moisture content also did not show any significant change during the second event. This indicates that the soil was already saturated and allowed the maximum of water reaching the bottom of the snowpack to appear as runoff. The water balance analyses of all rainfall events are given in Table 2. No loss of water in the form of infiltration to the soil is considered because of its saturation condition. Approximately the total input to the plots in the form of rainfall and snowmelt emerged as runoff from both snow plots during the second rainfall event. The snowpack became fully conditioned during this event because of sufficient rainfall input and therefore, most of the rainwater infiltrated through the snowpack Table 2 Details of total input from various sources to the snow plot for different events and observed runoff. All figures correspond to the unit area of the plot. Rain event Rain input Natural Rain-induced Total liquid water Observed (mm) snowmelt (mm) snowmelt (mm) input (mm) runoff (mm) Event 1 Î55TÔ TO CK9 1571) 51 ' Event

6 94 Pratap Singh et al. very quickly to appear as runoff. Any heavy rainfall with snowmelt can result in high streamflow causing floods under such conditions. To understand the storage behaviour of the snowpack, the liquid water-holding capacity of the snow was measured after the second rainfall event. These observations were made after the drainage of excess gravitational water from the snowpack and can be considered as representative of the liquid water-holding capacity. The average liquid water-holding capacity of snow without impeding effect was 6.6%, but it increased to about 14.4% near the ice surface. These high values may be because of the high saturation of the snowpack, impermeability of the ice layers and lack of sufficient slope which impeded horizontal drainage over the ice layers and water stayed in the snowpack. It indicates that the impeding characteristics of the ice layers increased the water-holding capacity of the snow more than 2 times and enhanced the water storage capacity substantially. This stage of the snowpack played an important role in reducing the risk of high streamflow due to heavy rainfall. CONCLUSIONS The field investigations describe the storage and transmission characteristics of water through snow for rain-on-snow occurrence and provides quantitative experimental data obtained from field investigations in the Austrian Alpine region. These investigations were carried out at plot-scale by simulating artificial rainfall with different intensity and duration, and studying the behaviour of the emerging outflow. The study reveals that the average liquid water-holding capacity of the snowpack without impeding effect of water was about 6.8%, but it increased to about 14.2% near the ice layers due to additional water impounded on their relatively impermeable surfaces. It shows that heterogeneities in the snow increased the water storage and retention capacity of the snowpack about 2 times and this stage of snowpack played an important role in reducing the risk of floods due to heavy rainfall over the snowpack. Under fully saturated conditions of the snowpack, the speed of water movement under heavy rainy conditions was observed to be about 6 m h" 1. The changes in metamorphism in the saturated snow and existence of preferential paths due to differential snow settlement during the rainy conditions are understood to be the main factors to produce water percolation with high speed. These vertical preferential flow paths tend to reduce the transient storage and travel time of water in the snowpack in comparison to that which would occur if uniform melting is taking place on the surface. This indicates that heavy rainwater moves several times faster than the natural snowmelt under non-rainy conditions. Moreover, under rainy conditions natural snowmelt also percolates faster along with rainwater enhancing the probability of flood. Most of the total water input appeared as runoff because negligible loss occurred from the input water flux. This stage of the snowpack is responsible for generating floods either from heavy rain alone or in combination with snowmelt. The results indicate that the role of rainwater is much more important for the conditioning of a snowpack to yield maximum runoff rather than contributing to additional melting of snow. Even heavy rainfall simulations could not produce significant rain-induced melt, but they saturated and conditioned the snowpack quickly to produce maximum and fast runoff.

7 The role ofsnowpack in producing floods under heavy rainfall 95 Acknowledgements The Austrian Academic Exchange Program is acknowledged for providing financial support for the stay of the principal author in Austria to carry out this work. REFERENCES Archer, D. R., Bailey J. O., Barrett, E. C. & Greenhill, D. (1994) The potential of satellite remote sensing of snow over Great Britain in relation to cloud cover. Nordic Hydrol. 25, Bergman, J. A. (1987) Rain-on-snow and soil mass failure in the Sierra Nevada of California. Landslide Activity in the Sierra Nevada during 1982 and Earth Resources Monograph 12, USDA Forest Service, Pacific Southwest Region, San Francisco, California, USA, Brunengo, M. J. (1990) A method of modelling the frequency characteristics of daily snow amount for stochastic simulation of rain-on-snowmelt events. In: Western Snow Conf. Proc. 58, Christner, J. & Harr, R. D. (1982) Peak streamflows from the transient snow zone, Western Cascades, Oregon. In: Proc. Western Snow Conf. Proc. 50, Colbeck, S. C. (1972) A theory of water percolation in snow. /. Glaciol. 11, Colbeck, S. C. (1974) Water flow through snow overlying an impermeable boundary. Wat. Resour. Res. 10, Colbeck, S. C. & Davidson, G. (1973) Water percolation through homogeneous snow. In: Role of Ice and Snow in Hydrology (Proc. Banff Symp., September 1972), IAHS Publ. no Conway, H. & Raymond, C. F. (1993) Snow stability during rain. J. Glaciol. 39, Conway, H., Breyfogle, S. & Wilbour C. R. (1988) Observations relating to wet snow stability. In: International Snow Science Workshop, ISSW'88 Comm. (Whistler, British Columbia, Canada). Dunne, T., Price, A. G. & Colbeck, S. C. (1976) The generation of runoff from subarctic snowpacks. Wat. Resour. Res. 12, Harr, R. D. (1986) Effects of clearcutting on rain-on-snow runoff in western Oregon: a new look at old studies. Wat. Resour. Res. 22, Kattelmann, R. C. (1987) Water release from a forest snowpack during rainfall. In: Forest Hydrology and Watershed Management (ed. by R. H. Swanson, P. Y. Bernier & P. D. Woodard) (Proc. Vancouver Symp., August 1987) IAHS Publ. no Sandersen, F., Bakkehai, S., Hestnes, E. & Lied, K. (1996) The influence of meteorological factors on the initiation of debris flows, rockfalls, rockslides and rockmass stability. Report no , Norwegian Geotechnical Institute, Oslo. Singh, P. & Kumar, N. (1997) Effect of orography on precipitation in the western Himalayan region. /. Hydrol. 199, Singh, P., Ramasastri, K. S. & Kumar, N. (1995) Topographical influence on precipitation distribution in different ranges of western Himalayas. Nordic Hydrol. 26, Wankiewicz, A. (1976) Water percolation within a deep snowpack field investigations at a site on Mt Seymour, British Columbia. PhD Thesis, University of British Columbia, Canada.