Dye tracer experiment

In this section, the observations from the dye tracer experiment at Hannigalp are summarized. The stained water flow paths observed in most of the excavated profiles showed a pronounced heterogeneous pattern. Distinct preferential flow fingers formed at the soil surface and led down to the somewhat coarser soil layer at 40-80 cm, just above the bedrock, where the water was able to spread. According to our observations, the hydrophobic soil surface, small ant channels and plant

Winter 2000/01 (without soil frost) Winter 2001/02 (with soil frost)

Winter 2000/01 (without soil frost) Winter 2001/02 (with soil frost)

Figure 7.8 Depth profiles of areal coverage of pixels stained with dye tracer for all profiles excavated on two dates of spring 2001 and 2002 at Hannigalp: one date shortly after the start, and the other during the final phase of the snowmelt. Consecutive profiles from the same date superimposed on each other

Areal coverage of dye tracer (-) Areal coverage of dye tracer (-)

Figure 7.8 Depth profiles of areal coverage of pixels stained with dye tracer for all profiles excavated on two dates of spring 2001 and 2002 at Hannigalp: one date shortly after the start, and the other during the final phase of the snowmelt. Consecutive profiles from the same date superimposed on each other

Figure 7.9 (a) Image of a soil profile excavated on March 16, 2002, showing the infiltrated meltwater concentrated in the uppermost 25 cm above frozen soil. (b) During spring 2002, a thin ice layer formed on the soil surface because of melting and refreezing

Figure 7.9 (a) Image of a soil profile excavated on March 16, 2002, showing the infiltrated meltwater concentrated in the uppermost 25 cm above frozen soil. (b) During spring 2002, a thin ice layer formed on the soil surface because of melting and refreezing roots were main generators of the preferential water flow. Not more than 50% of the uppermost soil layers were pervaded in the winter2000/01 (Figure 7.8). During that first winter (without soil frost), we did not notice indications for impeded water flow at any depth. Toward the end of the snowmelt, the stained infiltration front penetrated further down in the profile.

In the second winter (with massive soil frost down to a depth of 50 cm), the first snowmelt produced a slightly different infiltration pattern: the stained water concentrated in the top 25 cm, impeded by the underlying frozen soil layer (Figure 7.9). However, in the course of the spring, the stained water percolated downward although the soil frost persisted to the end of the snowmelt. In a similar fashion to the previous year, preferential flow paths formed in the upper part of the profile and spread in the coarser subsoil material. Although the stained profiles clearly showed considerable infiltration and percolation through the frozen soil, we also discovered considerable amounts of stained water 10 to 20 m downslope of the experimental site originating from lateral surface runoff. A closer examination confirmed that the thin ice layer on the soil surface (cf. previous section) had triggered that surface runoff. With our tracer experiment, we were able to confirm this layer in spite of its relatively thin (<5cm) extent (Figure 7.9).

To sum up, an impeding effect of the frozen soil was observed only during the first phase of the snowmelt. For the specific soil at Hannigalp, nevertheless, most of the meltwater was able to infiltrate into the frozen soil. More impeding than the frozen soil itself was the basal ice layer that built up on the surface.


With the calibrated COUP-model, we simulated a period of 30 years for each elevation zone. The simulated snow depth at 1600 m was compared to the measured one at Grachen (Figure 7.10). The snow depth was mostly satisfactorily reproduced by the model (coefficient of determination R2 = 0.84). In particular, onset and end of the snow period were well timed. The main differences were observed in January and February, when the snow depth was mostly underestimated by the model. Reasons for this error are twofold. On the one hand, the applied daily resolution in the meteorological inputs underestimates the snowmelt, because of the smoothing of the temperature gradient between the atmosphere and the snow during the daytime. On the other, the solar radiation at Grachen, which was not directly measured but estimated from the cloud cover, may have been underestimated in winter, because of the complex mountain topography, and the relatively low position of the sun in the sky at that latitude.

On the basis of these model simulations, we classified each winter as ''frozen'' (considerable soil frost during the snowmelt in >80% of the area), ''partially frozen'' or ''unfrozen'' (no soil frost during the snowmelt in >80% of the area). Since the beginning of the water table elevation measurements in 1992, three winters were ''frozen'' (1996 = winter 1995/96, 1998 and 2002), five winters ''unfrozen'' (1993,1995,1997,2000,2001), and two winters ''partially frozen'' (1994, 1999).

For each winter, we compared the water-table rise during the snowmelt with the accumulated winter precipitation (expressed as the areal average snow water equivalent at the start of the snowmelt) (Figure 7.11). The large water-table rise of 8 to 12 m normally starting in April and finishing by the end of June may be regarded as an indicator of aquifer recharge, neglecting any effect induced by the geology or the topography on the recharge.

The rise was lowest for the three ''frozen'' winters (less than 8 m). These winters were characterized by a shallow snowpack during the whole winter, resulting in a deep soil frost at each altitudinal zone. For the ''unfrozen'' winters, the rise was considerable (>10m), even when comparatively little winter precipitation was recorded, as in winter 1996/97. Especially interesting is the comparison between the two ''extreme'' winters 1997 and 2002. They were characterized by contrasting precipitation distribution, explaining the differences in the thermal soil state at snowmelt. In 1997, an early and thick snowpack prevented the soil from freezing in November and December, whereas the precipitation between January and May remained far below average.

- Measured - Simulated

; ;












' 1 ;

i J



i i

L, , J

1968 1970 1972 1974 1976 1978 1980 1982 1984


\ |






J L; i,...L.t

,1 • ;

4 I

..I... I





j 1 i , .1 i'l

1984 1986 1988 1990 1992 1994 1996 1998 2000

Figure 7.10 Measured and simulated snow depth at Grachen (1600 m) for the period 1968 to 2000

1984 1986 1988 1990 1992 1994 1996 1998 2000

Figure 7.10 Measured and simulated snow depth at Grachen (1600 m) for the period 1968 to 2000

o Unfrozen a Partially frozen Frozen



-T 1909





i ~T~ i I



4 f, Li 1...

...I 7X". ...I



1 I




210 260 310 360 410 460 510

210 260 310 360 410 460 510

Precipitation (mm)

Figure 7.11 Water-table rise during the snowmelt at Grachen versus winter precipitation (i.e areal average snow water equivalent at the start of the snowmelt) shown for each spring from 1993 to 2002

On the other hand, the relatively high amount of winter precipitation in 2002 was mainly caused by a large snowfall in May 2002, when more than 130mm of precipitation fell within four days. Despite significant differences in the winter precipitation, the water-table rise was 31% lower during the frozen winter 2002 than during the unfrozen winter 1997. Such a result may indicate that a partly frozen soil influences the snowmelt discharge over large areas. However, this interpretation should be viewed with caution because, on the one hand, little is known about the hydraulic behavior of the catchment, and, on the other, the error in the winter precipitation is large, because of the strong local variability in the precipitation.

The two ''partially frozen'' winters 1994 and 1999 illustrate the difficulty in accurately modeling the frost depth aerial extension when, at the onset of the winter, large variation in the snow depth exists. Indeed, despite a similar frost extension (the two lowest zones, between 1600 and 2000 m, were simulated as frozen, and the higher areas as unfrozen) both winters were characterized by contrasting effects on the water-table rise. In spring 1994, the rise was less than 10 m despite considerable snowfall, in contrast to spring 1999, when it was greatest (17.25 m).


The results from Hannigalp and Gd St Bernard corroborate the very sensitive relation between snow cover and soil frost. A shallow snowpack enables the ground to freeze deeply, whereas a thick snowpack may insulate the ground preventing soil frost - even at such high altitudes. From the simulation results, we noted that the occurrence of frost on the two experimental fields was sporadic and depended on the late autumnal and early winter weather conditions. So it is not surprising that during the last 10 years, a deep soil frost was encountered during three winters only at Hannigalp.

A frozen soil may influence the snowmelt discharge pattern considerably, as shown by the dye tracer experiment and the plot discharge measurements. Under frozen soil conditions, the penetration of the infiltrating wetting front was delayed, compared to unfrozen conditions. With regard to the lateral runoff, an increase from nil (unfrozen winters) to approximately 35% (frozen winters) of the total meltwater was observed. This drastic change was mainly caused by the presence of a sheet of ice at the base of the snowpack. When this basal ice layer had disappeared, most meltwater infiltrated into the ground, as shown by the late snowmelt event in May 2002 at Hannigalp. We believe that the formation of the basal ice sheet is favored by the rather cold mean soil temperature, the long snow cover period and the early snowmelt events, as the snow cover period is long enough to allow a substantial latent heat transfer between the wet basal snowpack and the upper frozen soil boundary. However, further investigations on the formation of basal ice-layers are needed. In sub-alpine areas, where the snow cover periods are shorter and snowmelt more intense, the presence of a basal ice sheet seems to be rare. In such areas, the soil ice content, as well as refreezing of snowmelt in the soil pores are the major factors that influence the amount of lateral runoff (Stadler etal. 1996).

During the main snowmelt period, we observed that a part of the recorded surface water infiltrated the soil some 100 m below the experimental plot where the soil was already free of snow and unfrozen. Such a result demonstrates the importance of the soil texture, structure and steepness, as well as the underlying geological structure on the amount of surface runoff. Although the soil texture was similar at both experimental sites, considerably more lateral runoff was measured at Gd St

Bernard, as the experimental plot was located on a much steeper slope than at Hannigalp.

Finally, in Grachen we noted the water-table rise at snowmelt was reduced by 10 -30% during frozen winters. This decrease is less marked than the groundwater recharge at Hannigalp, where, from simulation results, the deep seepage diminished between 20 and 50% of the total meltwater during frozen winters. The very permeable soil allowed most meltwater to re-infiltrate the soil in lower areas where frost was absent. These results support other studies showing that the effect of seasonal frost on the water circulation diminishes with increasing areal extension of the studied field (Thorne et al. 1998; Cherkauer and Lettenmaier 1999).

The following statements sum up our experiments and simulations.

• For the development or absence of soil frost both thickness and timing of the snow cover are decisive.

• Surface runoff depends largely on the presence of an ice layer at the base of the snowpack and the amount of soil moisture at the onset of the winter.

• A frozen soil considerably influences the discharge during snowmelt periods at the local scale.

• At larger scales, however, a considerable portion of meltwater is able to infiltrate the unfrozen ground somewhere downslope, due to spatial variability of the soil frost, of the hydraulic soil properties and of the steepness of the slope.

• In very permeable soil, soil frost reduces the water-table rise in spring only marginally.

Generalizing these results, we conclude that despite a massive snow cover, deep soil frost forms during specific winters at these altitudes, influencing the degree of groundwater recharge. A change in the discharge patterns due to seasonal frost may have relevant implications for the general water circulation, particularly with respect to flooding. During rain on snow events, the soil infiltration capacity is further reduced by the presence of frost. It results in an acceleration of the outflow from the snowpack, which in turn increases the amount of surface runoff, hence potentially increasing the risk for flooding.


Burt TP, Williams PJ (1976) Hydraulic conductivity in frozen soils. Earth SurfProc 1: 349-360. Cherkauer KA, Lettenmaier DP (1999) Hydrologic effects of frozen soils in the upper Mississippi River basin. J Geophys Res 104(D16): 19,599-19,610.

De Gaetano AT, Wilks, SS, McKay M (1996) A physically based model of soil freezing in humid climates using air temperatures and snow cover data. J Appl Meteorol, 35: 1009-1027.

Flerchinger GN, Saxton KE (1989) Simultaneous heat and water model of a freezing snow-residue-soil system I. Theory and development. Trans ASAE32(2): 565-571.

Forrerl, Papritz A, Kasteel R, FluhlerH, LucaD (2000) Quantifying dye tracers in soil profiles by image processing. Eur J Soil Sci 51(2): 313-322.

Harlan, RL (1973) Analysis of coupled heat-fluid transport in partially frozen soil. Water Resour Res 9: 1314-1323.

Ippisch O (2001) Coupled transport in natural porous media. Dissertation, University of Heidelberg, Germany, p. 145.

Jansson P-E, Halldin S (1979) Model for the annual water and heat flow in a layered soil. In: Halldin S (ed) Comparison of Forest and Energy Exchange Models. International Society for Ecological Modelling, Copenhagen, pp. 145-163.

Jansson P-E, Karlberg L (2001) Coupled heat and mass transfer model for soil-plant-atmosphere systems. TRITA-AMI Report 30 87, ISSN 1400-1306, KTH Stockholm, Sweden.

Johnsson H, Lundin L-C (1991) Surface runoff and soil water percolation as affected by snow and soil frost. JHydrol 122: 141-159.

Koren VI, Duan QI, Schaake JC (1995) Modeling ofthe effect of frozen ground on snowmelt/rainfall processes. International GEWEX workshop on cold season/region hydrometeorology. Banff, Alberta, pp. 78-82.

Marty C (2001) Surface radiation, cloud forcing and greenhouse effect in the Alps. Dissertation, ETHZ Institute for Climate Research, p. 122.

SeyfriedMS, MurdockMD (1997) Use of air permeability to estimate infiltrability of frozen soil. J Hydrol 202: 95-107.

Shanley JB, Chalmers A (1998) The effect of frozen soil on snowmelt runoff at sleepers river, Vermont Hydrol Process 13: 1843-1857.

Stadler D, Briindl M, Wunderli H, Auckenthaler A, Fluhler H (1996) Field measurements of water transport in frozen soils. Hydrol Process 10(10): 1293-1304.

Stadler D, Stahli M, Aeby P, Fluhler H (2000) Dye tracing and image analysis for quantifying water infiltration into frozen soils. Soil Sci Soc Am J 64: 505 -516.

Stahli M, Bayard D, WydlerH, FluhlerH (2004) Snowmelt infiltration into alpine soils visualized by dye tracer technique. Arct Antarct Alp Res 36(1): 128-135.

Stahli M, Jansson P-E, Lundin L-C (1999) Soil moisture redistribution and infiltration in frozen sandy soils. Water Resour Res 35: 95-103.

Stahli M, Nyberg L, Mellander P-E, Jansson P-E, Bishop KH (2001) Soil frost effects on soil water and runoff dynamics along a boreal transect: 2. Simulations. Hydrol Process 15: 927-941.

Thorne GA, Laporte J, Clarke D (1998) The effect of frozen soils on groundwater recharge and discharge in Granitic rock terrane of the Canadian shield. Nord Hydrol 29: 371-384.

0 0

Post a comment