Hydrology in Mountainous Regions. J - Hydrological Measurements; the Water Cycle (Proceedings of two Lausanne Symposia, August 1990). IAHS Publ. no. 193, 1990.

A lysimetric pillow station at Kùhtai/Tyrol

R. KIRNBAUER & G. BLÔSCHL Institut fur Hydraulik, Gewâsserkunde u. Wasserwirtschaft, Technische Universitât Wien, Karlsplatz 13, 1040 Vienna, Austria

ABSTRACT For properly forecasting snowmeIt-runoff the understanding of processes associated with a melting snow cover may be of primary importance. For this purpose a snow monitoring station was installed at Kuhtai/Tyrol at an elevation of 1930 m a.s.l. In order to study individual physical processes typical snow cover situations are examined. These situations include cold and wet snow under varying weather conditions. Based on a few examples the diversity of phenomena occuring at the snow surface and within the snow cover is demonstrated.

INTRODUCTION

Within a short-term flood-forecasting system a snowmelt model should be capable of representing extreme conditions. As Leavesley (1989) points out, a more physically based understanding of the processes involved will improve forecast capabilities. Subjective watching of phenomena together with measuring adequate data of sufficient accuracy and time resolution may form the foundations of process understanding. Most field studies performed so far concentrated on investigating the energy input to snow, particularly under melting conditions (see e.g. Kuusisto, 1986). Differences in the relative importance of processes during contrasting weather conditions have been reported by numerous authors (e.g. Lang, 1986). Considering these differences some of the authors (e.g. Anderson, 1973) distinguished between advection and radiation melt situations in their models. In this study meteorological data and snowpack observations from an alpine experimental plot are presented. Following an approach adopted earlier within the context of a model sensitivity analysis (Blôschl et âL-< 1988), it is believed that individual processes appear more explicitly in typical situations. Therefore, the investi­ gations presented here focus on selected periods of characteristic weather and snow cover conditions.

THE KÛHTAI EXPERIMENTAL PLOT

The snow monitoring station was set up on the site of the Kiihtai meteorological station near the Langental reservoir located about 30 km west of Innsbruck at an elevation of 1930 m a.s.l. (see Fig. 1). The research station site was graded prior to the installation of the instruments. Thus the site is flat with a slight slope to the west. It

173 R. Kirnbauer& G. Blôschl Il A

FIG. 1 The lysimetric station at Kuhtai/Tyrol (for abbreviations see Table 1).

is situated within the high relief environment of the Austrian Alps. In December, therefore, the station is barely met by the sun. The experimental plot is surrounded by typical timber-line vegetation with Alpine roses, meadows and scattered cembra-pines. The climate is characterized by mean annual precipitation of about 1100 mm, 45% of which fall as snow. The snow cover period typically starts in November and lasts until May. Maximum snow depths of about 150 to 200 cm are reached in March. The instrumentation comprises meteorological devices as shown in Fig. 1 and listed in Tab. 1. and a lysimetric snow pillow. The design of the lysimetric snow pillow is based on a device described by Engelen et. a]^. (1984). The hexagonal rubber pillow of 10m2 in area, filled with antifreeze, is connected to a stand pipe with a floating 175 A fysimetric snow pillow station atKiihtai/Tyrol

TABLE 1 Instrumentation of the Kiihtai experimental plot.

Variable Instrument Abbrev. in Fig. 1 global radiation (Schenk) RG reflected s.w. radiation pyranometer (Eppley) RR net radiation (Swissteco) RN air temperature resistance TA humidity capacitive sensor HA wind speed cup WS wind direction WD precipitation raingauge PG precipitation recording raingauge, heated, PR tipping bucket type snow temperatures; at resistance thermometer TS 8 levels above ground snow depth snow stake SS snow depth ultrasonic device SU water equivalent snow pillow & PI stand pipe SP melt rates snow lysimeter & LY tipping bucket TB

gauge. The meltwater draining the pillow is collected in a gutter at the edge of the lysimeter and measured by a tipping bucket of 0.05 mm resolution. Lateral inflow to the lysimeter is prevented by a drainage surrounding the device and a 20 cm metal lip. Cross checking of lysim­ eter and pillow data indicated that lateral inflow was negligible. During two weeks in April 1989 additional observations of snowpack characteristics were made. These included profiles of snow tempera­ ture and liquid water content at intervals of three hours. Snow temperatures were measured by a thermistor. Liquid water content was measured by a capacitive probe designed by Denoth & Foglar (1985). Each time at least two profiles were examined for assessing the horizontal variability of snow cover parameters. Under most conditions this variability was small as compared to temporal fluctuations.

SELECTED PERIODS OF TYPICAL WEATHER AND SNOW COVER CONDITIONS

Mid winter - low humidity conditions

As an example of cold snowpack conditions a period in December 1987 is presented in Fig. 2. It comprises bright and overcast days, horizon shading being most obvious on clear-sky days such as 22 and 24 December. Solar radiation clearly affected the diurnal variations of air temperature. At the beginning of the period humidity gradually decreased to a minimum of 1 mbar vapour pressure equivalent to 20% relative humidity. The snowpack first was at 0°C, subsequently cooled down and only returned to an isothermal state on 27 December. Snow R. Kirnbauer& G. Blôschl ne

GLOBAL 400 " RADIATION (W m-2) 300 200" 100 A I A .A. AIR 8 TEMPERATURE CO 4" M. 0 .V-VW ^v . ^y Av-"^- 4 VAPOUR B PRESSURE (mbar) 2 SNOW 0 TEMPERATURES CO

DEC.22 DEC.23 DEC.24 DEC.25 DEC.26 DEC.27

FIG. 2 Meteorological and snow temperature data measured during a mid winter period in 1987. temperatures were monitored at 10 and 30 cm above ground, the upper sensor being approximately 10 cm below surface. As may be observed from Fig. 2 the overall cycle of snow temperature at 30 cm is similar to that of air temperature. However, there are tendencies and day-to-day variations (e.g. 23-25 December) which do not correspond to air temperature. These variations indicate the influence of additional variables such as cloudiness and humidity. These influencing factors control surface energy exchange in terms of long wave radiation and latent heat flux. Particularly on fair weather days such as 22 and 24 December one may identify a substantial loss of energy due to long wave radiation. Prevailing evaporation conditions on days of very low humidity (e.g. 24 December) are assumed to significantly accelerate the cooling down of the snowpack. The considerable differences between air and snow temperatures observed during the major part of the period analyzed may be attributed to the combined effect of long wave emission and evaporation.

Melting period conditions

Fundamentally different phenomena may be observed during the melting period. Fig. 3 summarizes results of a 23 day period in April 1989. In this period the snowpack was ripe and experienced night time refreezing at the surface. Snowdepths varied at around 70 cm. Net radiation (in terms of potential melt rates) and lysimeter outflow are compared. For Fig. 3 the daily sum of net radiation is evaluated by integrating hourly values between 1900h and 1900h of the previous day considering the effect of night time radiative loss on the state of the snowpack in the morning. The daily sum of lysimeter outflow is 177 A lysimetric snow pillow station at Kiihtai/Tyrol

FIG. 3 Melting period 7-29 April 1989. Accumulative lysimeter outflow and radiation on a daily basis. based on the interval between 1200h and 1200h of the following day due to the well known fact that the melt wave lags behind energy input. Fig. 3 indicates that cumulative net radiation approximately equals cumulative meltwater outflow over the period considered. Hence one may infère that overall net radiation dominates melting processes. However, this is not true of individual days. Smaller and larger slopes of the graph in Fig. 3 indicate gain and loss of energy due to turbulent fluxes respectively. In the sequel four days within the period are discussed in more detail.

Overcast and clear sky conditions Two contrasting weather conditions are presented in Fig. 4. 13 April represents a typical rainy day whereas 24 April was a fair weather day. On 13 April changes in air temperature and net radiation were small. Humidity was near saturation and approximately 3 mm of mixed rain and snow were observed. The snowpack started draining in the afternoon yielding a maximum melt rate of 2 mm h-1. On 24 April there were pronounced fluctuations in air temperature, low humidity and substantial radiation loss during the night. The melt intensity was about half of that observed on the overcast day. In analyzing the differences in melt rates one may first look at Fig. 3. The graph shows that the daily net radiation values on 13 and 24 April were 7 and 8 mm whereas the daily melt rates were 12 and 7 mm respectively. On 13 April positive air temperatures along with high wind speeds indicate substantial melt due to sensible heat. On 24 R. Kirnbauer& G. Blôschl

AIR 8 TEMPERATURE -^V 0 V

-4

VAPOUR 6" PRESSURE (near) 4 2

WIND 4 SPEED (m s-1) 2 ^->/ VA-

NET 300 RADIATION (W m-2) 200 100" /v 0.

PRECIP. \ g" («s h-1) 0.8 „ „ mrJln LYSIMETER 3" OUTFLO* (mm h-1) 2™ f mm^^Jllihr Jk Oh 12h 24h Oh 12h 24h APRIL 13 APRIL 24

FIG. 4 Comparison of a clear sky and a rainy day in April 1989.

April, however, evaporative heat loss dominated over sensible heat in­ put resulting in a net heat loss due to turbulent fluxes. Additional­ ly, some of the differences may be attributed to (a) rain and (b) dif­ ferences in the state of the snowpack induced by preceding processes.

Drop of air temperature Fig. 5 illustrates the processes interrupting spring melt. The most striking feature of the weather situation on 26 and 27 April was a sharp drop of temperature during the night (Fig. 5 a). Low values of cloudiness were observed on the first day considered. Subsequently clouds were advected in the early morning of the second day causing overcast conditions on 27 April. A meltwave was produced on the warm and clear day, the trailing limb of which extended to the following cold day. In order to analyze the processes in more detail snow moisture profiles are presented. Fig. 5 b indicates that in the morning of 26 April the snow surface was frozen in spite of high air temperature and wind speed. This may be attributed to an energy loss due to longwave radiation. The interior of the snowpack was wet. Surface melt, produced between sunrise and HOOh, penetrated the snowpack down to a level of 40 cm above ground (shaded area in Fig. 5 b). The lysimeter hydrograph indicates that the meltwave reached the ground at 1300h. This statement is endorsed by the moisture profile taken at 1400h which shows high liquid water content due to transient water. During 179 A tysimetric snow pillow station at Kiihtai/Tyrol

AIR a TEMPERATURE b) (°c) 4- APRIL 26 APRIL 27

0

-4

VAPOUR PRESSURE (mbar)

WIND SPEED (m s-1)

NET 300 RADIATION (H a-2) 200 100"

LYSIHETER 3 OUTFLOW 20 40 60 80 20 40 60 80 (mm h-1) 2 WATER C0NTENT(kg m"3) WATER CONTEHT(kg m-3)

Oh 12h LOh )2h 24h APRIL 26 APRIL 27

FIG. 5 Drop of temperature on 26 and 27 April 1989. a) Meteorological data and lysimeter outflow. b) Liquid water content profiles.

the night melting ceased, and in the morning of 27 April the moisture profile was nearly identical to that of the previous day. This admits the inference that on both mornings the snow was saturated.

CONCLUSIONS

The diversity of phenomena associated with snowmelt in a high alpine environment has been illustrated on the basis of measurements at the Kuhtai experimental plot. In order to study individual physical processes typical meteorological and snow cover conditions are examined. During a period of low humidity in mid winter the snowpack cools down to temperatures considerably below air temperature due to longwave radiation and evaporation. During the ablation period snowmelt is dominated by net radiation. Data evidence indicates that turbulent fluxes may significantly increase and decrease melt on rainy and clear sky days respectively. A sharp drop of temperature in the late melt season results in freezing the snow surface but does not affect the moisture profile within the snowpack. If the concept of a snowmelt model is based on physical principles adequate simulation of such phenomena can be expected. R. Kirnbauer& G. Blôschl 180

ACKNOWLEDGEMENTS This research was supported by grants from the Austrian Fonds zur Fôrderung der wissenschaftlichen Forschung under project No. P7002PHY and from the Tyrolean Hydro-electric Power Company (TIWAG). Additionally to the monetary grants the TIWAG pro­ vided invaluable support in completing their Kiihtai meteorological station with the equipment necessary for snow monitoring. Staff of the Federal Institute for Snow and Avalanche Research (EISLF), Switzerland assisted in designing the snow temperature monitoring device.

REFERENCES

Anderson, E. A. (1973) National Weather Service river forecast system - snow accumulation and ablation model. NOAA Tech. Memo. NWS HYDR0-17, US Dept. of Commerce, Silver Soring, Maryland. Blôschl, G., Gutknecht, D. & Kirnbauer, R. (1988) Berechnung des Wârmeeintrages in eine Schneedecke - Analyse des Einflusses unterschiedlicher meteorologischer Bedingungen (Simulation of heat input to snow - analysis of the influence of different météorologie conditions). Deutsche Gewâsserkundliche Mitteilunqen 32 (1/2), 34-39. Denoth, A. & Foglar, A. (1985) Measurements of daily variations in the subsurface wetness gradient. Annals of Glacioloqy 6., 254-255. Engelen, G. B., van de Griend, A. A. & Valentini, P. (1984) A lysimetric snow-pillow station for continuous monitoring of the snow cover cycle and its processes at the "Seiser Aim", South Tyrol, N. Italy. In: Schneehydroloqische Forschung in Mitteleuropa. 129-143. Mitteilungen des Deutschen Verbandes fur Wasserwirtschaft und Kulturbau e.V. no.7. Kuusisto, E. (1986) The energy balance of a melting snow cover in different environments. In: Modelling SnowmeIt-Induced Processes (Proc. Budapest Symposium, July 1986), 37-45. IAHS Publ. no. 155. Lang, H. (1986) Forecasting meltwater runoff from snow-covered areas and from glacier basins. Chapter 5 in: River Flow Modelling and Forecasting (eds. Kraijenhoff, D. A. & Moll, J. R.). D. Reidel Publishing Company, Dordrecht. Leavesley, G. H. (1989) Problems of snowmelt runoff modelling for a variety of physiographic and climatic conditions. Hydroloqical Sciences - Journal - des Sciences Hydroloqigues 34 (6), 617-634.