1. Introduction
Pomeroy et al. (1998) report that interception by forest canopies can store up to 60% of the cumulative snowfall, resulting in a 30%–40% annual loss of snow cover in boreal coniferous forest environments. Because of the large surface-area-to-mass ratio of the coniferous canopy, and the fact that snow remaining in the canopy is exposed to a comparably dry and warm atmosphere, high rates of sublimation can occur, which are largely dependent on the residence time of intercepted snow. For an accurate determination of exposure times, it is important to know the amount of snow intercepted. Generally, snow interception efficiency increases with canopy density and the available deposition surface, increasing size of falling snow crystals, decreasing density of the falling snow, decreasing temperature, and decreasing wind speed (Marsh 1999; Pomeroy and Gray 1995). According to field measurements, the capacity of the canopy interception storage is on the order of 10 mm (Hedstrom and Pomeroy 1998; Pomeroy and Gray 1995) (Fig. 1).
(left) Intercepted snow on Alpine fir trees one day after a heavy snowfall. (right) Inside-canopy snow interception on the branches of sub-Alpine firs. Both sites in the Bavarian Alps (Germany). Photos: U. Strasser.
Citation: Journal of Hydrometeorology 12, 4; 10.1175/2011JHM1344.1
In recent years, numerous models have been developed and applied to quantify the amount of snow intercepted in evergreen forest canopies and the rates at which the intercepted snow sublimates (Pomeroy et al. 1998; Hedstrom and Pomeroy 1998; Link and Marks 1999a,b; Pomeroy et al. 2002; Essery et al. 2003; Liston and Elder 2006). Only the remainder of intercepted snow, which does not undergo sublimation, can melt or unload from the canopy and thus contribute to the snow stored on the ground. Models used to simulate snow–canopy interactions comprise parameterizations of the interception and sublimation processes, and depend on meteorological forcing and parameters describing tree characteristics. The large surface-area-to-mass ratio and frequently high wind speeds in mountain forest canopies enable efficient turbulent exchange that promotes sublimation. The resulting proportion of snowfall that sublimates varies depending on interception efficiency, exposure time, and atmospheric conditions.
In addition to interception and sublimation, micrometeorological conditions at the surface of a ground snow cover are modified by the existence of a forest canopy. These include decreases in shortwave radiation, precipitation, wind speed, and the diurnal amplitude of air temperature, and increases in longwave radiation and humidity (Link and Marks 1999a,b; Tribbeck et al. 2004; Strasser et al. 2008). Consequently, latent fluxes, and the associated sublimation rates at the snow surface below a forest canopy, differ from those in the open.
Strasser and Etchevers (2005) have shown that consideration of meteorological conditions at the ground beneath a canopy significantly improves the simulated amount and timing of meltwater release and the hydrological modeling of a high Alpine catchment. Results from the distributed simulations of Link and Marks (1999a,b) indicate that canopy structure can delay seasonal snowmelt in densely forested areas by up to three weeks relative to open areas.
Generally, the effect of snow losses due to interception and sublimation yields declining snow accumulation beneath the forest canopy in response to increasing canopy density and leaf area (Pomeroy et al. 2002). In contrast, differences in micrometeorological conditions in the forest canopy tend to delay the snowmelt in springtime mainly because of reduced incoming solar radiation.
The interaction between these processes and their opposing impact on the seasonal evolution of snow cover are only partly understood. Currently, snow scientists are unable to consistently predict snow accumulation and ablation beneath the canopy relative to nearby open sites. The numerical experiment described herein is intended to enable the analysis of these processes and effects under idealized conditions. The main objective is to test drive the different processes represented within a detailed, physically based, snow–canopy model using simplified terrain and realistic meteorological data. For this purpose, we constructed an idealized, cone-shaped mountain covered with a geometrically regular pattern of coniferous forest stands and clearings. The lack of surrounding terrain and its effect on thermal radiation, shading, etc.; the regularly shaped topography of the mountain; and the known distribution of forests and clearings simplifies the analyses. The meteorological data are recorded datasets from two stations for which both the real horizontal and vertical distances are preserved in the simulation domain, thus ensuring physical plausibility of the produced meteorological and snow-process fields.
2. Methods
a. Data
Input data required for distributed snow processes modeling consists of a digital elevation model (DEM), the sky view factor (SVF) describing the portion of the visible hemisphere at each location, canopy distribution and structure, and hourly meteorological recordings of precipitation, global radiation, temperature, humidity, and wind speed.
1) The idealized mountain
We created an idealized mountain landscape to conduct the numerical experiments. The geometry of the domain is visualized in Fig. 2. The resolution of the raster elements (pixels) is 50 m, corresponding to an area of 2500 m2 each. The spatial extent of our domain is 15 km with the mountain positioned in its center. The radius from the mountain’s foot to its summit is 4 km, or 80 pixels. The elevations of the mountain base and its top correspond to the actual elevations of the two weather stations used: 617 m MSL (Schoenau) and 1407 m MSL (Kuehroint). Hence, the mountain’s slope is 11.3° (20%). Kuehroint is positioned at the mountain top to preserve the physical meaning of the recorded data and the lapse rates. The valley station Schoenau is “multiplied” and distributed around the mountain four times, maintaining the real-world distance between Kuehroint and Schoenau (5350 m). The mountain geometry and meteorological station locations are the same for all simulations. The sky view factor is computed from the DEM applying the algorithm of Corripio (2003). Finally, the forest canopy is distributed in a chessboard-like pattern of stands and clearings around the mountain. It is characterized by its effective leaf area index (LAIeff) including stems, leaves, and branches (Chen et al. 1997). The canopy distribution is illustrated in Fig. 2. Forest-covered areas and clearings are distributed over 5 elevation zones in sectors of 45°, ensuring close areas with and without forest in all altitudes and exposures. We chose LAIeff = 2, LAIeff = 8, and LAIeff = 14 to represent sparse, medium, and dense forest stands, respectively. Canopy characteristics only vary between the different model runs and are held constant spatially.
Setup of the numerical experiment: (top) the modeling domain and (bottom) the idealized mountain with the meteorological stations located on top of it and around the base. In the bottom right panel, shading represents forested areas.
Citation: Journal of Hydrometeorology 12, 4; 10.1175/2011JHM1344.1
2) Meteorological data
The meteorological datasets utilized in our study are provided by the operational network of automatic meteorological stations located in Berchtesgaden National Park. The two stations Kuehroint and Schoenau are part of the infrastructure maintained by the Bavarian Avalanche Warning Service (LWZ; www.lawinenwarndienst-bayern.de/) of the State Office for Environment (station Kuehroint) and the German Weather Service (DWD; www.dwd.de/) (station Schoenau). Figure 2 indicates the location of the two stations in the Berchtesgaden National Park and the position of our idealized mountain using white contour lines. Station Schoenau is located in a nonforested valley area (617 m MSL) and the station Kuehroint is located in a mountain pasture (1407 m MSL). Table 1 provides an overview of the variables recorded at each station. All data were sampled every 10 seconds and recorded every 10 minutes (i.e., average for temperature, humidity, wind speed, radiation, and air pressure; total for precipitation). Recordings were then aggregated to hourly values.
Altitude, location, and set of recorded parameters with temporal resolution for the meteorological stations of the automatic network in the Berchtesgaden National Park used in this investigation. Coordinate system is Universal Transverse Mercator (UTM). Accuracies of the recordings used in the modeling is 0.3 m s−1 (wind speed), ±0.3°C (temperature), ±1% (humidity), ±5% (global radiation), and <0.4% (precipitation). Accuracy of the snow pillow recording SWE is 0.25%, and is 0.1% for the ultrasonic ranger measuring snow depth. All accuracies are from the technical specifications provided by the manufacturers. Parameters are wind speed (WS), wind direction (WD), temperature (T), temperature at 0.05 m (T0.05), humidity (H), snow depth (SD), SWE, sunshine duration (SS), global radiation (GR), direct radiation (DR), reflected radiation (RR), precipitation (P), and air pressure at sea level (AP).
In our experiment, we used station data from three seasons: 2005/06, 2006/07, and 2007/08, each from August to July the following year. Table 2 shows mean values of air temperature and sums for precipitation for the three years and the winter seasons from November to May.
Temperature means and precipitation sums for the 2 meteorological stations during the 3 winter seasons 2005/06, 2006/07, and 2007/08.
Measurements of snow water equivalent (SWE) at station Kuehroint were collected using a snow pillow, and snow depth was measured with an ultrasonic ranger. They show the evolution of snow cover for the three modeled winter seasons (Fig. 3). Here 2005/06 is characterized by continuous accumulation starting in December, with maximum SWE at the end of March and a late snowfall period from mid- to late April. This leads to a snow-rich winter with a mean SWE of 310 mm and a mean snow depth of 1004 mm during November–May. In contrast, 2006/07 represents a winter with an exceptionally small snow amount (mean SWE: 36 mm, mean snow depth: 198 mm). The snow cover melts off completely several times during midwinter, which is atypical at this altitude for the studied area. Finally, 2007/08 starts with an early snowfall event in September and a typical snow accumulation from November to February. Mean values of SWE and snow depth (mean SWE: 281 mm, mean snow depth: 869 mm) indicate a winter similar to the 2005/06 season. To conclude, our meteorological forcing data enables us to simulate three winter seasons, two of which represent conditions typical for the region, while one is atypical with only very small snow amounts.
Seasonal snow-cover evolution at the Kuehroint station site (1407 m MSL) as modeled and recorded with a snow pillow.
Citation: Journal of Hydrometeorology 12, 4; 10.1175/2011JHM1344.1
b. Model
The model applied in this study, the Alpine Multiscale Numerical Distributed Simulation Engine (AMUNDSEN) (Strasser et al. 2004, 2008), is a simulation framework for continuous, distributed modeling of snow processes in high mountain areas. The functionality of AMUNDSEN includes (i) rapid computation of topographic parameters from a digital elevation model, (ii) several interpolation routines for scattered meteorological measurements (Strasser et al. 2004), (iii) simulation of shortwave and longwave radiative fluxes including consideration of shadows and cloudiness (Corripio 2003; Greuell et al. 1997), (iv) parameterization of snow albedo (U.S. Army Corps of Engineers 1956; Rohrer 1992), (v) modeling of snowmelt with either an energy balance model (Strasser et al. 2008) or an enhanced temperature index model considering radiation and albedo (Pellicciotti et al. 2005), (vi) modeling of forest snow processes (Liston and Elder 2006), and (vii) simulation of snow slides from steep mountain slopes along couloir courses derived from the DEM (Gruber 2007). The latter is required to remove snow from areas where it would otherwise infinitely accumulate in long-term simulations; this is not used in the simulations described herein.
AMUNDSEN simulations of mountain snow-cover evolution have been compared to measurements in a variety of geographical regions by several investigators (e.g., Strasser et al. 2002; Zappa et al. 2003; Strasser et al. 2008), including analyses at regional scales using satellite data–derived snow cover (Strasser and Mauser 2001; Prasch et al. 2007). As an example, comparison of modeled plot-scale SWE with snow pillow recordings from the three winter seasons investigated herein are illustrated in Fig. 3 for the Kuehroint station site.
1) Beneath-canopy climate
Presence of a forest canopy changes the micrometeorological conditions at the surface of the ground snow cover below. Shortwave radiation, precipitation, and wind speed are reduced, whereas longwave radiation and humidity are increased and the diurnal temperature cycle is attenuated (Strasser and Etchevers 2005; Tribbeck et al. 2004; Link and Marks 1999a,b). In AMUNDSEN, the below-canopy modifications of the micrometeorological conditions over the ground snow surface are modeled explicitly. The meteorological station data (solar and thermal radiation, temperature, humidity, and wind speed) that were collected in the open are interpolated over the domain and modified for forest canopies. These meteorological distributions are then used to drive AMUNDSEN and its snow–canopy submodels. The term LAIeff is the primary forest parameter and represents the only stand characteristic required for the modeling.
The inside-canopy modification of meteorological parameters has been measured and the respective parameterization applied and validated with good results by Durot (1999) for the Col de Porte observation site (1420 m MSL) in the French Alps.
2) Ground snow processes
3) Canopy–snow processes and forest canopy model
Various modeling approaches have been developed to simulate snow–vegetation interactions (see Hedstrom and Pomeroy 1998, Pomeroy et al. 1998, Hardy et al. 2000, and Pomeroy and Gray 1995 for reviews). In AMUNDSEN, the processes of interception, sublimation, unloading by melt, and exceeding the canopy snow–holding capacity are calculated as a function of LAIeff using well-documented parameterizations from the literature, mostly following the scheme of Liston and Elder (2006).
Reported modeled seasonal intercepted snow sublimation ranges from 13% of annual snowfall for a mixed spruce–aspen, to 31% for mature pine, and 40% for a mature spruce stand in the southern boreal forest of central Canada (Waskesiu Lake, Prince Albert National Park, Saskatchewan) (Pomeroy et al. 1998). Measurements of sublimation from intercepted snow within a subalpine forest canopy at a U.S. continental site amounted to 20%–30% of total snowfall accumulated at the site (Montesi et al. 2004).
The snow interception and sublimation model implemented in AMUNDSEN applies our physical understanding of snow interception at branch and canopy scales, and scales the corresponding understanding of snow sublimation of a single snow crystal to the intercepted snow within the canopy. When canopy air temperatures are above freezing, intercepted snow is melted and transferred to the ground store. Snow unloading by wind is a complex process that depends on canopy structure and spacing, branch and trunk flexibility, snow temperature and precipitation histories, and wind speed and direction (Liston and Elder 2006). Numerical quantification of these processes awaits a field study not yet implemented.
Apart from sublimation, snow can also be removed from the interception store by melt unloading. The snow masses are hereby assumed to fall down to the ground or snowpack below after a partial melt at the surface. Melt unload Lm (kg m−2) is estimated for temperatures above freezing using the temperature index melt model of Pellicciotti et al. (2005). For the estimation of the load of snow falling to the ground, the following scheme is applied. Using field observations, Liston and Elder (2006) estimated a daily unloading rate of 5 kg m−2 day−1 K−1. By applying a scaling factor of 3.3, the temperature index melt model was calibrated to fit this estimate. The resulting unloaded mass is calculated in each time step and added to the ground snow cover beneath the trees.
By means of this snow–canopy interaction model, the processes of interception, snow sublimation, and melt unload are simulated. In a period of heavy snowfall, the interception store can be filled to its maximum capacity. From the interception store, snow is removed by sublimation and melt unloading induced by a period of positive temperatures.
Both the simulated rates of sublimation and the combined effect of melt and release of intercepted snow strongly depend on effective LAI, which is the only canopy characteristic or parameter used in the model—it modifies canopy transmissivity for solar radiation, wind speed, canopy density, and the maximum interception storage capacity. Strasser et al. (2008) quantified the sensitivity of the model on effective LAI for processes affecting both canopy-intercepted and ground snow. The snow interception model has been validated by Montesi et al. (2004) using observations from a continental climate site located within the USDA Fraser Experimental Forest (39°53′N, 105°54′W) near Fraser, Colorado. Integrated in AMUNDSEN, the scheme was compared to a variety of other snow–canopy interaction parameterizations within the framework of the Snow Models Intercomparison Project-2 (SnowMIP2) and showed good results (Rutter et al. 2009).
3. Results and discussion
To explore and understand snow–canopy interactions on mountain slopes, the simulated patterns from the numerical experiments are discussed. The model results vary in space because of altitude, exposure, radiative fluxes, and interpolated meteorological variables (Strasser et al. 2004). Where not stated otherwise, effective LAI in the figures is 14 to best visualize the difference in snow processes between canopy and open areas.
a. Seasonal snow-cover evolution
Seasonal ground snow-cover evolution is strongly affected by the existence of a forest canopy. Figure 4 shows the modeled snow water equivalent at selected points in the model domain during the 2005/06 season. We choose four points, all at the same elevation (1307 m MSL) and 100 m below the summit of the idealized mountain. Two of them are situated on the northern side (N) and two on the southern side (S) of the mountain. Each “exposure pair” consists of a forested (LAIeff = 8) and an adjacent open site.
Modeled snow water equivalent at selected points in the model domain during the 2005/06 snow season.
Citation: Journal of Hydrometeorology 12, 4; 10.1175/2011JHM1344.1
The development of SWE at the two nonforested points (P1 and P3) clearly shows the influence of the exposure on the snow cover. During the accumulation period in December and January, there are no significant differences between N and S. However, when it comes to losses of SWE due to melt and sublimation, the ablation at N is considerably slower than at S as a result of the larger incoming solar radiation. This leads to less SWE from January on, an earlier snow-free date, and shorter snow-cover duration at S compared to N. Maximum SWE values occur at the end of March, with up to 450 mm at the S and 500 mm at the N site. In contrast, the development of SWE at the two forested points (P2 and P4) is almost identical. The shading of incoming shortwave solar radiation by the canopy causes the snow cover to develop almost completely independent of exposure. For LAIeff = 2, we found slight differences of about 30 mm SWE between the two exposures related to the effects described above. The maximum SWE on the ground beneath the canopy is lower than at the nonforested points, reaching about 350 mm in mid-April for both N and S exposures. This delay is caused by temporarily intercepted snow that later falls to the ground and late snowfall that melts in the open but not in the forest, whereas the lower absolute value can be explained by sublimation losses of intercepted snow.
During the accumulation period from December to March there is less snow beneath the canopy compared to the open areas, independent of exposure. This gap increases with time because of the constant accumulation of snow losses due to canopy interception and sublimation. When melting events become predominant from March on, the gap gets smaller until the SWE is equal in the open and beneath the forest at the end of March at S and in mid-April at N. From this time on, the relations invert and there is more snow beneath the canopy than in the adjacent open areas; consequently, the melt-free date is delayed.
Generally, a canopy conserves the snow on the ground beneath it because of shading of incoming solar radiation during melting periods and, in contrast, reduces snow reaching the ground during accumulation periods. The nearly uniform snowpack growth during winter of 2005/06 leads to a clear effect of these processes in reduced snow masses on the ground in the early and midwinter months and an extended beneath-canopy snow-cover duration in springtime. The following winter season (2006/07) shows that, with a more uneven snow-cover evolution, the described processes can lead to different and more complex results (Fig. 5). The snow cover melts off totally several times because of sustained ablation periods during midwinter. The effect of reduced snow precipitation reaching the ground beneath the canopy during the accumulation periods is similar to the previous winter season. But looking at the differences between the forested and open sites during the ablation events, the season and exposure dependency of the canopy influence becomes visible. During the early ablation events in October and November, the diminishing melt impact of the canopy due to radiation shading is negligible. Therefore, the snowpack melts off earlier beneath the canopy. The ablation periods from December to January first show a snow-conserving effect for the S exposed slope in the middle of December where snow melts off simultaneously at the forested and open sites. The next three ablation phases impressively indicate the increasing influence of the canopy-shading effect on snowmelt as the season progresses. In the beginning of February, snowmelt beneath the canopy is retained. At the N slope, this leads to a melt off at the same point in time at the forested and open sites despite less snow precipitation having reached the ground beneath the canopy. For the S reference points, the snow cover beneath the canopy has a longer duration than in the open. This trend continues throughout the ablation events in the spring months with a considerable extension of snow-cover duration beneath the canopy. The increasing snow-conservation effect of the canopy can be explained by an increase of the energy input fraction for snow ablation that comes from solar radiation as the season progresses. For the same reason—higher incoming shortwave radiation—the effect of the snowpack being protected from melt by the canopy is much more pronounced on the S slopes compared to the N.
Modeled snow water equivalent at selected points in the model domain during the 2006/07 snow season.
Citation: Journal of Hydrometeorology 12, 4; 10.1175/2011JHM1344.1
Altogether, the effect of snow losses due to canopy interception and sublimation significantly influences the snow-cover buildup during accumulation periods. A reduced amount of snow precipitation reaches the ground, which results in a lower maximum SWE beneath the canopy compared to the adjacent open area. The snow-conserving effect of the forest canopy depends on exposure and season. It becomes remarkable from February on, with a maximum on southern slopes, and becomes more and more dominant throughout the spring melt season.
b. Snowfall and snow-cover duration
Figure 6 (first row) illustrates the accumulated snowfall reaching the ground in the open and at the ground beneath the canopy during the three modeled winter periods. The maximum snowfall amounts of 910.5 mm (2007/08) and 765.6 mm (2005/06) in the two regular winter periods once more demonstrate the extraordinary small snowfall amounts in 2006/07 (412.7 mm). This is confirmed again by the mean values of snow precipitation reaching the ground over the idealized mountain, which are approximately 450 mm in the two regular winters and only 217.6 mm in winter 2006/07. These and additional minimum, maximum, and mean values of the relevant model results are listed in Table 3.
(top) Snowfall at the ground, (second row) snow sublimation from the canopy, (third row) snow-cover duration, and (bottom) snowmelt for the 3 winter seasons (left) 2005/06, (center) 2006/07, and (right) 2007/08.
Citation: Journal of Hydrometeorology 12, 4; 10.1175/2011JHM1344.1
Main characteristics of snow-cover evolution for the 3 winter seasons 2005/06, 2006/07, and 2007/08.
The three seasons indicate consistent spatial patterns of snowfall, which are a result of two superimposed subpatterns. Snowfall amounts increase with altitude because of decreasing air temperatures and increasing precipitation as a result of the measured station data and the applied interpolation method. The second pattern is the arrangement of forested and nonforested areas on the idealized mountain. Looking at a specific elevation, there is always more ground snow accumulation in the open areas compared to the adjacent forested ones. The snow that is temporarily intercepted by the canopy and that later falls to the ground is included in the accumulated snowfall. This fact implies that the differences in ground snowfall between forested and open areas are solely a result of the high sublimation rates from snow intercepted in the canopy into the atmosphere, compared to the small mass losses due to sublimation from the ground snowpack in both the open areas and the snow surface beneath the canopy.
Snow-cover duration (third row in Fig. 6) during a winter season is the result of all processes determining the evolution of the snowpack, including mechanisms that affect both snow accumulation and ablation. Main drivers for these processes are the prevailing meteorological conditions. Generally, the number of days with a snowpack existing on the ground increases with elevation because of rising snow precipitation and decreasing air temperatures. The snow-cover duration ranges from 48 days in winter 2006/07 at the nonforested and lowest elevations of site S to 242 days at the highest elevation forested areas in winter 2007/08 (Table 3).
Despite more snowfall reaching the ground in the open areas, in most cases the snow-cover duration in those areas is shorter than the adjacent forested regions. This is a result of the radiation shading effect of the canopy described above. An exception can be seen in the winter season 2006/07. In contrast to the regular pattern, at the highest elevation zones on the N slopes of the mountain (see the points marked in Fig. 6), the nonforested area has a slightly longer snow-cover duration than the forested one. This can be explained by the uneven snow-cover evolution in winter 2006/07 and the season- and exposure-dependent canopy influence. Snow ablation events in early winter lead to a relatively large discrepancy in SWE between nonforested S and N slopes because of larger amounts of snow remaining longer on the N slopes. The canopy interception and sublimation process produces a large gap between N forested and N open sites during the early winter accumulation period. At the N forested site, the small snow amounts melt off quickly and from there on there is no snow to be protected by the forest canopy (e.g., at end of December, visualized in Fig. 5). In this special case, snow-cover duration is therefore larger at the N open site than at the N forested site. In other words, a prerequisite for the protective effect of the forest canopy is an existing snowpack that can be protected from melt.
At the S slope, the protective influence is effective during the spring months and, as a consequence, the snow-reducing effect is compensated by the protective one, leading to a generally prolonged snow-cover duration beneath the canopy (Figs. 4 and 6).
In addition, the nonforested areas show differences in snow-cover duration that are dependent on exposure, as expected. Because of higher incoming solar radiation and therefore higher amounts of energy available for melt and sublimation, the snow-cover duration is shortened at S slopes compared to other exposures at the same elevation. The protective impact of the canopy reduces these exposure-dependent differences under the forest canopy, until ultimately, if the forest density is great enough, the differences are eliminated.
c. Snow sublimation
Seasonally accumulated snow sublimation of canopy-intercepted snow is an important component in the Alpine winter water balance (Strasser et al. 2008). Generally, snow sublimation from both the ground as well as from the trees of a canopy increases with the availability of radiative energy, more efficient turbulence, and lower humidity, leading to a higher saturation deficit. In analyzing total seasonal amounts, the overall duration of the snow period itself plays a vital role. Figure 6 (second row) clearly shows the increase of seasonal snow sublimation quantities from the canopy with increasing elevation for all three winter seasons. However, maximum values are quite different, ranging from 100 mm (2006/07) to almost 250 mm (2005/06). Such large sublimation totals require high energetic input, which can only occur when large, rough snow surfaces are exposed to high radiation and efficient turbulence for a long time, as is the case for the idealized mountain forest in 2005/06.
The spatial pattern of snow sublimation from the ground mostly depends on snow availability (i.e., the duration of snow coverage) and is therefore similar to the patterns in Fig. 6—in snow-rich winters (2005/06) there is a relatively small difference of snow-cover duration between S and N and, hence, total sublimation amounts are similar in all exposures. For winters with short snow-cover duration, however, the difference in snow-cover duration between S and N becomes larger along with the difference in snow sublimation amount. Nevertheless, total amounts of ground sublimation only reach values of a few mm per season (a maximum of 23.7 mm in our study). Inside the canopy, values are even smaller, mainly because of less radiation (Eq. 1), higher humidity (Eq. 6), and reduced wind speed (Eq. 7). Modeled resublimation is in a similar range (Table 3), so the balance of snow-to-moisture conversion and vice versa at the ground surface becomes almost zero. Both sublimation and resublimation patterns are discussed and visualized in detail by Strasser et al. (2008).
d. Snowmelt
Accumulated snowmelt during a winter season (Fig. 6, bottom) is largely determined by the total snowfall reaching the ground (Fig. 6, top). Hence, the amount of snowmelt increases with elevation. In addition, the sublimation of snow held within the canopy is much greater than the sublimation from the nonforested snowpack (Table 3). Furthermore, there are also small sublimation rates from the snowpack beneath the canopy. This leads to smaller snow sublimation losses and therefore higher accumulated snowmelt rates in the open compared to an adjacent forested area. An exception can be found in the lowest elevation zones during the winter seasons 2006/07 and 2007/08, where snowmelt amounts are slightly higher beneath the canopy than on the open ground. This is a result of snow-cover duration in the respective seasons together with the model system definitions. During melting periods, the snow is conserved beneath the forest canopy. In spring, the lower-elevation, open areas are free of snow from a certain time on, whereas the ground beneath the canopy is still covered by a wet snowpack. Within the modeling system, liquid precipitation passes through the canopy and falls down to the ground or onto the snow cover below. Rain that falls on the snowpack contributes to later snowmelt amounts. In contrast, rain that falls on bare ground does not contribute to snowmelt. In that special case, “trapping” of liquid precipitation, combined with increased snow-cover duration, leads to larger snowmelt amounts beneath the canopy. The values of seasonal snowmelt range from 234 to 1045 mm. Minimum, maximum, and mean values for the simulated winter seasons are listed in Table 3.
4. Conclusions
The virtual numerical experiment presented herein allows a depiction of the effects of snow–canopy interactions on the patterns of snow processes in a mountain forest. Both the rates and seasonal sublimation totals of intercepted snow in a canopy are much higher than sublimation losses from the snow surface at the ground. This leads to significantly reduced snow accumulation on the ground beneath a canopy. However, shadowing and reduced radiative input under the trees leads to a process of protection and, because of reduced melt and sublimation, the below-canopy snow-cover duration is extended compared to that in the open. The effect of reduced accumulation is dominant during winter, whereas the shadowing effect with reduced ablation prevails during spring. In addition, the specific meteorological conditions and the general evolution of the winter season play an important role, as shown by the occurrences in winter 2006/07 when only a shallow and short-lasting snow cover developed—the canopy prolongs the ground snow cover compared to the open snow cover only for the case of a minimum amount of snow. If only a small amount of snow exists, the effect of reduced radiation at the forest ground alone cannot overbalance the reduced accumulation, and the snow cover disappears earlier than in the open.
Furthermore, the experiment shows that snow hydrological processes inside a canopy only slightly depend on exposure. In the open, snow processes contain effects that vary much more, depending on which side of the mountain the slope is situated.
The overall goal of our study was to simplify the setting or model setup in such a way that the influencing forces can be separated and understood relative to their effect on spatial snow patterns and the respective differences between a forest and an open snow cover. The regular, chessboard-like pattern of forest stands and open areas enabled elaboration of the differences between the two snow cover types. However, because of the complexity of the relevant processes, it is still difficult to depict and separate the causes of some details in the resulting snow patterns. Here, the model could still be improved; for example, melt and release of snow from the trees are still computed using a simple temperature index approach; shadowing at the timberline is not considered; resublimation of moisture onto the surface of needles, branches, or the stem are not considered; wind-induced transport of snow with its complex effects on accumulation into clearings were not considered; conversion of shortwave canopy heating to thermal radiation is not yet explicitly modeled; etc.
Nevertheless, this study allows us to derive the following set of simple rules of how a mountain canopy modifies the evolution of the seasonal snow cover:
A forest canopy represents an interception storage for snowfall, and the amount of snow that sublimates does not reach the ground. This results in reduced snow accumulation beneath the canopy.
A forest canopy produces a shadowing effect beneath the trees, leading to reduced radiative energy reaching the ground.
Exposure effects on the snow cover and its evolution are significant in the open but small in the canopy, and become important only for small effective LAIs.
The difference in snow-cover duration between a forest snow cover and an open snow cover are more significant for southern exposures.
In snow-rich winters, the shadowing effect is dominant and snow lasts longer inside the forest than in the open.
In winters with little snow, snow sublimation losses are dominant and, consequently, the snow lasts longer in the open than inside the forest, mainly for northern exposures.
In early and high winter, the radiation protection effect of shadowing by the canopy is small. If little snow is available, an intermittent melt out of the snow cover inside the forest can occur. In late winter and spring, the shadowing effect becomes more efficient and the canopy develops its protective value.
Acknowledgments
We are thankful to all who provided their help and assistance with writing this manuscript; in particular, Helmut Franz and Michael Vogel from the National Park Administration (Berchtesgaden) helped with data processing and subsidizing our activities. The meteorological data were generously supplied by the Administration Union of the Berchtesgaden-Koenigssee region in Berchtesgaden (Schoenau) and the Bavarian Avalanche Warning Center of the State Office for Environment in Munich (Kuehroint). Specifically, we acknowledge the valuable comments of Robert Hellstrom (Bridgewater, Massachusetts) and an anonymous colleague in the review process. Printed with support from the University of Graz (Austria).
REFERENCES
Buck, A. L., 1981: New equations for computing vapor pressure and enhancement factor. J. Appl. Meteor., 20, 1527–1532.
Chen, J. M., Rich P. M. , Gower S. T. , Norman J. M. , and Plummer S. , 1997: Leaf area index of boreal forests: Theory, techniques, and measurements. J. Geophys. Res., 102, 29 429–29 443.
Cionco, R. M., 1978: Analysis of canopy index value for various canopy densities. Bound.-Layer Meteor., 15, 81–93.
Corripio, J. G., 2003: Vectorial algebra algorithms for calculating terrain parameters from DEMs and solar radiation modelling in mountainous terrain. Int. J. Geogr. Inf. Sci., 17, 1–23.
Durot, K., 1999: Modélisation hydrologique distribuée du bassin versant nivo–pluvial de Sarennes: Validation des données d’entrée et dévelopement d’un module de fonte nivale sous forêt. Ph.D. dissertation, Institut National Polytechnique de Grenoble, 332 pp.
Essery, R. L. H., Pomeroy J. W. , Parvianen J. , and Storck P. , 2003: Sublimation of snow from boreal forests in a climate model. J. Climate, 16, 1855–1864.
Fleagle, R. G., and Businger J. A. , 1980: An Introduction to Atmospheric Physics. Academic Press, 432 pp.
Greuell, W., Knap W. , and Smeets P. , 1997: Elevational changes in meteorological variables along a midlatitude glacier during summer. J. Geophys. Res., 102, 25 941–25 954.
Gruber, S., 2007: A mass-conserving fast algorithm to parameterize gravitational transport and deposition using digital elevation models. Water Resour. Res., 43, W06412, doi:10.1029/2006WR004868.
Hardy, J. P., Melloh R. , Robinson P. , and Jordan R. , 2000: Incorporating effects of forest litter in a snow process model. Hydrol. Processes, 14, 3227–3237.
Hedstrom, N. R., and Pomeroy J. W. , 1998: Measurements and modelling of snow interception in the boreal forest. Hydrol. Processes, 12, 1611–1625.
Hellström, R. Å., 2000: Forest cover algorithms for estimating meteorological forcing in a numerical snow model. Hydrol. Processes, 14, 3239–3256.
Lee, L. W., 1975: Sublimation of snow in a turbulent atmosphere. Ph.D. dissertation, University of Wyoming, 162 pp.
Link, T., and Marks D. , 1999a: Distributed simulation of snowcover mass- and energy-balance in the boreal forest. Hydrol. Processes, 13, 2439–2452.
Link, T., and Marks D. , 1999b: Point simulation of seasonal snow cover dynamics beneath boreal forest canopies. J. Geophys. Res., 104, 27 841–27 857.
Liston, G. E., and Elder K. , 2006: A distributed snow-evolution modeling system (SnowModel). J. Hydrometeor., 7, 1259–1276.
Marsh, P., 1999: Snowcover formation and melt: Recent advances and future prospects. Hydrol. Processes, 13, 2117–2134.
Montesi, J., Elder K. , Schmidt R. A. , and Davis R. E. , 2004: Sublimation of intercepted snow within a subalpine forest canopy at two elevations. J. Hydrometeor., 5, 763–773.
Obled, Ch., 1971: Modèle mathématique de la fusion nivale. Ph.D. dissertation, Institut de Mécanique de Grenoble, 170 pp.
Pellicciotti, F., Brock B. , Strasser U. , Burlando P. , Funk M. , and Corripio J. G. , 2005: An enhanced temperature-index glacier melt model including shortwave radiation balance: Development and testing for Haut Glacier d’Arolla, Switzerland. J. Glaciol., 51, 573–587.
Pomeroy, J. W., and Schmidt R. A. , 1993: The use of fractal geometry in modelling intercepted snow accumulation and sublimation. Proc. 50th Eastern Snow Conf., Quebec City, QC, Canada, ESC, 1–10.
Pomeroy, J. W., and Gray D. M. , 1995: Snowcover accumulation, relocation, and management. NHRI Science Rep. 7, 144 pp.
Pomeroy, J. W., Gray D. M. , Shook K. R. , Toth B. , Essery R. L. H. , Pietroniero A. , and Hedstrom N. , 1998: An evaluation of snow accumulation and ablation for land surface modelling. Hydrol. Processes, 12, 2339–2367.
Pomeroy, J. W., Gray D. M. , Hedstrom N. , and Janowicz J. R. , 2002: Prediction of seasonal snow accumulation in cold climate forests. Hydrol. Processes, 16, 3543–3558.
Prasch, M., Strasser U. , and Mauser W. , 2007: Validation of a physically based snow model for the simulation of the accumulation and ablation of snow (ESCIMO). Proceedings: Alpine Snow Workshop, National Park Berchtesgaden Rep. 53, 78–91.
Rohrer, M. B., 1992: Die schneedecke im Schweizer Alpenraum und ihre modellierung. Zürcher Geogr. Schr., 49, 170–178.
Rutter, N., and Coauthors, 2009: Evaluation of forest snow processes models (SnowMIP2). J. Geophys. Res., 114, D06111, doi:10.1029/2008JD011063.
Strasser, U., and Mauser W. , 2001: Modelling the spatial and temporal variations of the water balance for the Weser catchment 1965–1994. J. Hydrol., 254, 199–214.
Strasser, U., and Etchevers P. , 2005: Simulation of daily discharges for the upper Durance catchment (French Alps) using subgrid parameterization for topography and a forest canopy climate model. Hydrol. Processes, 19, 2361–2373.
Strasser, U., Etchevers P. , and Lejeune Y. , 2002: Intercomparison of two snow models with different complexity using data from an Alpine site. Nord. Hydrol., 33, 15–26.
Strasser, U., Corripio J. G. , Pellicciotti F. , Burlando P. , Brock B. , and Funk M. , 2004: Spatial and temporal variability of meteorological variables at Haut Glacier d’Arolla (Switzerland) during the ablation season 2001: Measurements and simulations. J. Geophys. Res., 109, D03103, doi:10.1029/2003JD003973.
Strasser, U., Bernhardt M. , Weber M. , Liston G. E. , and Mauser W. , 2008: Is snow sublimation important in the alpine water balance? Cryosphere, 2, 53–66.
Thorpe, A. D., and Mason B. J. , 1966: The evaporation of ice spheres and ice crystals. Br. J. Appl. Phys., 17, 541–548.
Tribbeck, M. J., Gurney R. , Morris E. M. , and Pearson D. W. C. , 2004: A new Snow-SVAT to simulate the accumulation and ablation of seasonal snow beneath a forest canopy. J. Glaciol., 50, 171–182.
U.S. Army Corps of Engineers, 1956: Snow hydrology: Summary report of the snow investigations. North Pacific Division Corps of Engineers, 437 pp.
Zappa, M., Pos F. , Strasser U. , Warmerdam P. , and Gurtz J. , 2003: Seasonal water balance of an Alpine catchment as evaluated by different methods for spatially distributed snowmelt modelling. Nord. Hydrol., 34, 179–202.
