• Adams, R. S., , Spittlehouse D. L. , , and Winkler R. D. , 1996: The effect of a canopy on the snowmelt energy balance. Proc. 64th Annual Meeting Western Snow Conf., Bend, OR, Western Snow Conference, 171–174.

  • Beaumont, R. T., 1965: Mt. Hood pressure pillow snow gauge. J. Appl. Meteor., 4 , 626631.

  • Betts, R. A., 2000: Offset of the potential carbon sink from boreal forestation by decreases in surface albedo. Nature, 408 , 187190.

  • Chen, J., , Rich P. W. , , Gower S. T. , , Norman J. M. , , and Plummer S. , 1997: Leaf area index of boreal forests: Theory, techniques and measurements. J. Geophys. Res., 102 , 2942929444.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Decagon Devices, 1991: Sunfleck PAR Ceptometer operator’s manual. Decagon Devices, Pullman, WA, 77 pp.

  • Golding, D. L., , and Swanson R. H. , 1986: Snow distribution patterns in clearings and adjacent forest. Water Resour. Res., 22 , 19311940.

  • Hanson, C. L., 2001: Long-term precipitation database, Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28312834.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hanson, C. L., , Marks D. , , and Van Vactor S. S. , 2001: Long-term climate database, Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28392841.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hanson, C. L., , Pierson F. B. , , and Johnson G. L. , 2004: Dual-gauge system for measuring precipitation: Historical development and use. J. Hydrol. Eng., 9 , 350359.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hardy, J. P., , and Hansen-Bristow K. J. , 1990: Temporal accumulation and ablation patterns of the seasonal snowpack in forests representing varying stages of growth. Proc. 58th Western Snow Conf., Sacramento, CA, Western Snow Conference, 23–34.

  • Hardy, J. P., , Davis R. E. , , Jordan R. , , Li X. , , Woodcock C. , , Ni W. , , and McKenzie J. C. , 1997: Snow ablation modeling at the stand scale in a boreal jack pine forest. J. Geophys. Res., 102 , 2939729405.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hardy, J. P., , Melloh R. , , Robinson P. , , and Jordan R. , 2000: Incorporating effects of forest litter in a snow process model. Hydrol. Processes, 14 , 32273237.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Henderson-Sellers, A., , and Wilson M. F. , 1983: Surface albedo data for climatic modeling. Rev. Geophys. Space Phys., 21 , 17431778.

  • Ji, Q., , and Tsay S. , 2000: On the dome effect of Eppley pyrgeometers and pyranometers. Geophys. Res. Lett., 27 , 971974.

  • Jordan, R., 1991: A one-dimensional temperature model for a snow cover: Technical documentation for SNTHERM. CRREL Tech. Rep. 91-16, 64 pp.

  • Jordan, R., 1992: Estimating turbulent transfer functions for use in energy balance models. CRREL Tech. Rep. 1107, 17 pp.

  • Koivusalo, H., , and Kokkonen T. , 2002: Snow processes in a forest clearing and in a coniferous forest. J. Hydrol., 262 , 145164.

  • Lehning, M., , Bartelt P. , , Brown B. , , Russi T. , , Stöckli U. , , and Zimmerli M. , 1999: SNOWPACK model calculations for avalanche warning based upon a new network of weather and snow stations. Cold Reg. Sci. Technol., 30 , 145157.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Link, T. E., , and Marks D. , 1999: Point simulation of seasonal snow cover dynamics beneath boreal forest canopies. J. Geophys. Res., 104 , 2784127857.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks, D., 1988: Climate, energy exchange, and snowmelt in Emerald Lake Watershed, Sierra Nevada. Ph.D. thesis, University of California, Santa Barbara, 158 pp.

  • Marks, D., , and Dozier J. , 1979: A clear-sky longwave radiation model for remote alpine areas. Arch. Meteor. Geophys. Bioklimatol., 27B , 159187.

    • Search Google Scholar
    • Export Citation
  • Marks, D., , and Winstral A. , 2001: Comparison of snow deposition, the snow cover energy balance, and snowmelt at two sites in a semiarid mountain basin. J. Hydrometeor., 2 , 213227.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks, D., , Cooley K. R. , , Robertson D. C. , , and Winstral A. , 2001: Long-term snow database, Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28352838.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marshall, S. E., , and Warren S. G. , 1987: Parameterization of snow albedo for climate models. Large Scale Effects of Seasonal Snow Cover, B. E. Goodison, R. G. Barry, and J. Dozier, Eds., IAHS Publication 166, 43–50.

    • Search Google Scholar
    • Export Citation
  • Marshall, S., , Roads J. O. , , and Glatzmaier G. , 1994: Snow hydrology in a general circulation model. J. Climate, 7 , 12511269.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pearson, D., , Daamen C. C. , , Gurney R. J. , , and Simmonds L. P. , 1999: Combined modelling of shortwave and thermal radiation for one-dimensional SVATs. Hydrol. Earth Syst. Sci., 3 , 1530.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pierce, L. L., , and Running S. W. , 1988: Rapid estimation of coniferous forest leaf area index using a portable integrating radiometer. Ecology, 69 , 17621767.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pomeroy, J. W., , Parviainen J. , , Hedstrom N. , , and Gray D. M. , 1998: Coupled modelling of forest snow interception and sublimation. Hydrol. Processes, 12 , 23172337.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Price, A. G., , and Dunne T. , 1976: Energy balance computations of snowmelt in a sub-arctic area. Water Resour. Res., 12 , 686694.

  • Roujean, J. L., 1996: A tractable physical model of shortwave radiation interception by vegetative canopies. J. Geophys. Res., 101 , 95239532.

  • Rouse, W. R., 1984: Microclimate at arctic tree line. 1. Radiation balance of tundra and forest. Water Resour. Res., 20 , 5766.

  • Shook, K., , and Gray D. M. , 1997: Snowmelt resulting from advection. Hydrol. Processes, 11 , 17251736.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Slaughter, C. W., , Marks D. , , Flerchinger G. N. , , Van Vactor S. S. , , and Burgess M. , 2001: Thirty-five years of research data collection at the Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28192823.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thomas, G., , and Rowntree P. R. , 1992: The boreal forests and climate. Quart. J. Roy. Meteor. Soc., 118 , 469497.

  • Tribbeck, M. J., 2002: Modelling the effect of vegetation on the seasonal snow cover. Ph.D. thesis, University of Reading, 240 pp.

  • Tribbeck, M. J., , Gurney R. J. , , Morris E. M. , , and Pearson D. W. C. , 2004: A new Snow-SVAT to simulate the accumulation and ablation of seasonal snowcover beneath a forest canopy. J. Glaciol., 50 , 171182.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Viterbo, P., , and Betts A. K. , 1999: The forecast impact of changes to the albedo of the boreal forests in the presence of snow. J. Geophys. Res., 104 , 2780327810.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wiscombe, W. J., , and Warren S. G. , 1980: A model for the spectral albedo of snow. I. Pure snow. J. Atmos. Sci., 37 , 27122733.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Flowchart description of SNOWCAN.

  • View in gallery

    Snow depth at an open, windblown site.

  • View in gallery

    Snow temperature, density, and grain size profiles at an open, windblown site on (a)–(c) day 122, (d)–(f) day 140, (g)–(i) day 167, and (j)–(l) day 199. Two sets of measurements were made on day 122, shown separately by different shading.

  • View in gallery

    Depth and mass of snow at an open, sheltered site.

  • View in gallery

    Comparison between observed and simulated snow depth beneath a fir canopy.

  • View in gallery

    Evaluation of snow temperature and density profiles beneath a fir canopy on (a)–(c) day 139 and (d)–(f) day 167.

  • View in gallery

    Continuous simulated and measured snow temperature at (a) 30, (b) 20, and (c) 10 cm above the snow–soil interface and (d) at the interface.

  • View in gallery

    Reduction of solar radiative flux beneath a fir canopy.

  • View in gallery

    Subcanopy solar radiation in a 5-day period: (a) simulated downwelling radiation compared to array-averaged measurements and (b) simulated upwelling radiation compared to a single radiometer.

  • View in gallery

    Subcanopy longwave radiation (a) downwelling and (b) upwelling.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 3 3 0
PDF Downloads 1 1 0

The Radiative Effect of a Fir Canopy on a Snowpack

View More View Less
  • 1 Environmental Systems Science Centre, University of Reading, Reading, United Kingdom
  • 2 British Antarctic Survey, Natural Environment Research Council, Cambridge, United Kingdom
© Get Permissions
Full access

Abstract

Models of snow processes in areas of possible large-scale change need to be site independent and physically based. Here, the accumulation and ablation of the seasonal snow cover beneath a fir canopy has been simulated with a new physically based snow–soil vegetation–atmosphere transfer scheme (Snow-SVAT) called SNOWCAN. The model was formulated by coupling a canopy optical and thermal radiation model to a physically based multilayer snow model. Simple representations of other forest effects were included. These include the reduction of wind speed and hence turbulent transfer beneath the canopy, sublimation of intercepted snow, and deposition of debris on the surface.

This paper tests this new modeling approach fully at a fir site within Reynolds Creek Experimental Watershed, Idaho. Model parameters were determined at an open site and subsequently applied to the fir site. SNOWCAN was evaluated using measurements of snow depth, subcanopy solar and thermal radiation, and snowpack profiles of temperature, density, and grain size. Simulations showed good agreement with observations (e.g., fir site snow depth was estimated over the season with r2 = 0.96), generally to within measurement error. However, the simulated temperature profiles were less accurate after a melt–freeze event, when the temperature discrepancy resulted from underestimation of the rate of liquid water flow and/or the rate of refreeze. This indicates both that the general modeling approach is applicable and that a still more complete representation of liquid water in the snowpack will be important.

* Current affiliation: Scott Polar Research Institute, Cambridge, United Kingdom

Corresponding author address: M. J. Tribbeck, Environmental Systems Science Centre, University of Reading, 3 Earley Gate, Reading RG6 6AL, United Kingdom. Email: mjt@mail.nerc-essc.ac.uk

Abstract

Models of snow processes in areas of possible large-scale change need to be site independent and physically based. Here, the accumulation and ablation of the seasonal snow cover beneath a fir canopy has been simulated with a new physically based snow–soil vegetation–atmosphere transfer scheme (Snow-SVAT) called SNOWCAN. The model was formulated by coupling a canopy optical and thermal radiation model to a physically based multilayer snow model. Simple representations of other forest effects were included. These include the reduction of wind speed and hence turbulent transfer beneath the canopy, sublimation of intercepted snow, and deposition of debris on the surface.

This paper tests this new modeling approach fully at a fir site within Reynolds Creek Experimental Watershed, Idaho. Model parameters were determined at an open site and subsequently applied to the fir site. SNOWCAN was evaluated using measurements of snow depth, subcanopy solar and thermal radiation, and snowpack profiles of temperature, density, and grain size. Simulations showed good agreement with observations (e.g., fir site snow depth was estimated over the season with r2 = 0.96), generally to within measurement error. However, the simulated temperature profiles were less accurate after a melt–freeze event, when the temperature discrepancy resulted from underestimation of the rate of liquid water flow and/or the rate of refreeze. This indicates both that the general modeling approach is applicable and that a still more complete representation of liquid water in the snowpack will be important.

* Current affiliation: Scott Polar Research Institute, Cambridge, United Kingdom

Corresponding author address: M. J. Tribbeck, Environmental Systems Science Centre, University of Reading, 3 Earley Gate, Reading RG6 6AL, United Kingdom. Email: mjt@mail.nerc-essc.ac.uk

1. Introduction

Seasonal snow cover provides a significant proportion of the annual water budget for many ecosystems. Knowledge of the timing and magnitude of melt water runoff enables this resource to be managed more efficiently. Forest cover affects the snow cover in a number of ways, but it is impossible to measure all these effects simultaneously throughout the winter season. Afforestation may increase the rate of snowmelt (Golding and Swanson 1986) or delay snowmelt relative to snow in open areas (Rouse 1984; Hardy and Hansen-Bristow 1990). It is not clear which is the dominant energy term that contributes to snowmelt beneath the forest, but it is generally accepted to be radiative, either solar (Hardy et al. 2000) or thermal (Adams et al. 1996). The relative contribution of the radiative components to the energy balance depends on the characteristics of the canopy. A physically based model of the environmental system is needed to diagnose the contributions of these effects for any given site and to determine the overall effect of the forest cover on the evolution of snow beneath it.

Parameterization of the snow cover has a crucial impact on the performance of general circulation models (Henderson-Sellers and Wilson 1983; Thomas and Rowntree 1992; Marshall et al. 1994). Representation of the effect of the forest cover on the accumulation and ablation of the snow beneath the canopy is particularly important, given the size of the boreal forest, which covers 15% of the earth’s land surface and where precipitation is in solid form for 6–8 months of the year.

Reduced surface albedo for snow-covered areas with an overlying forest canopy was found to reduce dramatically the 2-m temperature underpredictions by the European Centre for Medium-Range Weather Forecasts model (Viterbo and Betts 1999). Betts (2000) examined the effect of surface albedo change with forestation and found that the increase in radiative forcing may more than offset the negative forcing from carbon sequestration, and therefore contribute to climate warming. The need to incorporate vegetation dynamics in climate models was clearly demonstrated by Betts (2000).

A new snow–soil vegetation–atmosphere transfer scheme (Snow-SVAT) to investigate the effect of vegetation on the seasonal snow cover was described by Tribbeck et al. (2004). This model, SNOWCAN, was developed in order to understand the radiative interaction between forest canopies and snow, yet be computationally simple enough to apply over different areas without the need for extensive field measurements to initialize the model. The interaction of radiation between the snowpack and forest canopy is simulated by coupled physically based snow and canopy models. The snow component model is a multilayer one-dimensional model based on SNTHERM (Jordan 1991). The canopy component determines the scattering and absorption of solar and thermal radiation through the canopy, and multiple scattering between the canopy and snowpack is simulated. The model is designed to represent complex processes with physical models, which are then site independent. Once a model that represents all processes accurately exists, sensitivity analysis will be used to identify how such a model can be simplified. This will then allow us to develop simpler models that will still retain the ability to simulate the snowpack at different sites and under different conditions, with a good degree of accuracy, and without tuning the model. These simpler models will then be incorporated in coupled global and regional models, where improved physical representations are important, particularly for studies of global and regional change.

SNOWCAN has been tested with snow depth measurements beneath five different canopies, and with subcanopy solar radiation measurements (Tribbeck et al. 2004). Although that study indicated that an increase in canopy density increased early and midseason melt rates but decreased melt rates later in the season, the model had not been validated thoroughly. In this paper, an evaluation of SNOWCAN is presented for accumulation and ablation periods, with comparisons of simulated and observed subcanopy radiation, snow depth, and snowpack profiles of temperature, density, and grain size. Detailed observations such as these are necessary to evaluate whether the model physics reflects actual behavior. This more complete evaluation is necessary in order to see where the model represents processes well and where further work is required. This in turn can be used to design simpler but still physically based representations for general circulation and similar models.

2. Model description

a. Snow model

The snow component model is a physically based one-dimensional model that uses conservation of mass, energy, and momentum to simulate the change in physical state of the snowpack. The model is based on the physics represented in SNTHERM (Jordan 1991), with differences in the style and compaction parameterization, and has different precipitation, solar zenith angle, and numerical solution algorithms (Tribbeck et al. 2004). Multiple layers are represented in the model to capture the stratification of snow, and horizontal homogeneity is assumed within each layer. A Lagrangian adaptive grid, which compresses with the compacting snowpack, is used to represent the layers. New layers are created for fresh precipitation events, and layers are combined or subdivided to optimize computational efficiency. The number and size of layers within the model are therefore governed by the precipitation history and desired accuracy of the simulation.

Heat conduction and transport of liquid water (gravitational drainage) and water vapor (diffusion) between layers are simulated, in addition to phase change within the layers. The effect of grain growth on the absorption of optical radiation is also simulated, with the rate of increase of the effective grain diameter d given by Eq. (1) for dry snow, where Uυ is the vapor flux. For wet snow, the grain growth is given by Eq. (2), where θl is the volumetric liquid water content, and the maximum rate of grain growth for wet snow is reached at a liquid water content of 9%:
i1525-7541-7-5-880-e1
i1525-7541-7-5-880-e2
Densification of the snow cover results from changes in the snowpack structure (metamorphism) and pressure from overlying snow (overburden). These processes result in a change in snow layer thickness, Δz, given by Eq. (3). Densification due to metamorphism is assumed to occur more rapidly for fresh snow, and the densification rates are modified by a factor e−0.046(γiγcrit) above a critical density, γcrit, where γi is the partial density of ice. Wet-snow metamorphism is also more rapid than that for dry snow, and the densification by metamorphism is assumed to occur twice as fast as for dry snow. Densification rates are thus dependent on the snow temperature (T) and density (ρs). Here Ps is the overburden pressure. The parameters ξ and η0 may vary between sites; values used in Tribbeck et al. (2004) for snow beneath boreal forest canopies were applied here:
i1525-7541-7-5-880-e3
At the lower boundary, temperature is specified at a defined depth within the soil, whereas horizontal runoff of liquid water is assumed at the snow–soil interface. At the upper boundary, energy fluxes are specified at the snow surface [Eq. (4)], and condensation or sublimation from the surface is determined from atmospheric boundary layer approximations:
i1525-7541-7-5-880-e4
where subscripts SW, LW, sen, lat, and prcp represent solar and thermal radiation, turbulent transfer of sensible and latent heat, and the heat convected by precipitation, respectively. The radiation components are modified by the interaction between the canopy and snow surface, and are described in the canopy radiation model section. However, the snow albedo is presented here, given by Eqs. (5) and (6). As described by Hardy et al. (1997), this algorithm was based on work by Marks (1988), Wiscombe and Warren (1980), and Marshall and Warren (1987). The clear-sky albedos for visible (αvis) and near-infrared (αnir) are defined as
i1525-7541-7-5-880-e5
where r is the effective grain size radius (in meters), fdir is the fraction of solar radiation that is direct, and θs is the solar zenith angle. The broadband albedo is calculated from the fraction of visible light fvis and adjusted for litter effects by the fractional coverage of litter (Λ), given the litter albedo (αV):
i1525-7541-7-5-880-e6
The turbulent transfer of sensible and latent heat are given by Eqs. (7) and (8), respectively, where EH0 and EE0 are windless exchange coefficients determined empirically. The density and heat capacity of the air are defined as ρair and Cair, the wind speed as w, and Lυi is the latent heat of sublimation. Further, CH and CE are bulk transfer coefficients for sensible and latent heat, and the surface vapor pressure, Pυ,sat is defined relative to ice (dry snow) or liquid water (wet snow). The bulk transfer coefficients are not parameters—they vary throughout the season and depend on the atmospheric stability. A full description of the turbulent transfer algorithms is given in Jordan (1992):
i1525-7541-7-5-880-e7
i1525-7541-7-5-880-e8
A sensitivity study of SNOWCAN was carried out by Tribbeck (2002), who found that the model was more sensitive to measurement errors in the forcing data than to the parameters, although the need for further investigation of the densification parameterization was identified.

This snow model was coupled to a solar and thermal radiation canopy model to simulate the evolution of snow cover beneath the forest canopy. This approach was taken because the dominant effect of the forest canopy on the energy budget is radiative; less important effects have been parameterized.

b. Canopy radiation model

The canopy solar and thermal radiation model coupled to the physically based snow model is the radiative transfer model “RM” (Pearson et al. 1999). Canopy elements are approximated as a series of discrete and randomly orientated scatterers. Only three parameters are required to describe the scattering and absorption of radiation: leaf area index, canopy single-scatter albedo, and emissivity. Direct solar radiation is approximated as a Dirac-delta function in the direction of the sun, and the attenuation of direct solar radiation is dependent on the pathlength through the canopy. Diffuse solar radiation is assumed to be isotropic from the sky. Thermal emission from the canopy and snow surface is simulated with Stefan’s law. Canopy temperatures are required as input and the onus is on the user to decide how to source the data or approximate these temperatures. The transmission and reflection of thermal radiation from the sky, canopy, and snow surface are represented in a similar manner to diffuse solar radiation.

The underlying basis for this model is the assumption that the canopy elements have a greater probability of overlapping (other elements) as the canopy density increases. This means that the change in radiative scattering is decreased in denser canopies. Equation (9) defines the third exponential integral, used to model the probability of overlap; τυ is the vegetation optical depth, which is dependent on the leaf area index, AL:
i1525-7541-7-5-880-e9
In this implementation, a spherical distribution of canopy elements is assumed (τυ = AL/2). This assumption may not be totally valid for canopies with a high degree of clumping, which increases radiation transmission through the canopy (Roujean 1996). One alternative is the use of effective leaf area index, as optical measurements of leaf area index already include the effects of clumping (Decagon Devices 1991) and woody elements of the canopy (Chen et al. 1997). Other relations between optical depth and leaf area index may be used (if known) to reflect the degree of clumping within the canopy, but may not be necessary as RM is less sensitive to the structure of vegetation than might be expected (Pearson et al. 1999).
Radiative transfer theory enables scattering coefficient R and transmission coefficient T to be defined for solar and thermal radiation, as shown in Eqs. (10) and (11) (Pearson et al. 1999): solar radiation:
i1525-7541-7-5-880-e10
thermal radiation:
i1525-7541-7-5-880-e11
where αυ and ɛυ are the canopy single-scatter albedo and emissivity. Multiple scattering of radiation is simulated between the snow surface and canopy to determine total downwelling (12) and upwelling radiation (13), where B is the lower boundary reflectance; that is, snow albedo for solar radiation and (1 − snow emissivity) for thermal radiation and I0 is the above-canopy incident solar radiation:
i1525-7541-7-5-880-e12
i1525-7541-7-5-880-e13
Radiative effects of the canopy on the snow energy balance are dominant but, in addition to radiative effects, the forest cover has a number of other effects that must be considered for accurate simulation of the snow cover. These canopy effects are treated less rigorously than the radiation interaction and have been parameterized simply for this paper, whose aim it is to examine the radiative effects.

Wind speed below the canopy is assumed to be 20% of that in the open (above 1 m s−1). This algorithm has been applied by Link and Marks (1999) and is consistent with the measurements made by Price and Dunne (1976) and Koivusalo and Kokkonen (2002). A logarithmic wind profile is assumed within the canopy for the calculation of turbulent transfer, with the roughness length governed by the snow surface features. The models are thus coupled in a radiative sense but linked in terms of turbulent transfer. A constant proportion of precipitation is assumed to be intercepted on the canopy, and all intercepted snow is assumed to be sublimated. These simplifying assumptions are discussed later.

c. Formation of SNOWCAN

In SNOWCAN, meteorological data used to drive the model are measured above the canopy or at a nearby open site. The required forcing data are downwelling solar and thermal radiation, wind speed, air temperature, relative humidity, and precipitation snow water equivalent (SWE). Measured canopy temperatures are useful to determine the emission of thermal radiation from the canopy, but are assumed to be equal to the air temperature if unavailable.

An adaptive time step is used to interpolate the meteorological data. The size of the time step is a function of the accuracy of the solution, but may vary between 5 s and 15 min. Simulation of the snowpack beneath a forest canopy is shown schematically in Fig. 1. After interpolation of the meteorological data, subcanopy fluxes are determined and used to calculate the mass and energy balance of the snowpack. Snow albedo is determined as a result of the growth of snow grains, and is used in the next time step to determine subcanopy radiation fluxes. A full description of SNOWCAN is given by Tribbeck et al. (2004)

3. Site and data description

Fieldwork was carried out during winter 2000/01 within the Reynolds Mountain East catchment, which is a semiarid subbasin of the Reynolds Creek Experimental Watershed in the Owyhee Mountains of southwestern Idaho. This was a collaborative effort with the U.S. Department of Agriculture/Agricultural Research Service (USDA/ARS) Northwest Watershed Research Center, which maintains the experimental watershed. Reynolds Mountain East is a 0.36-km2 headwater basin that contains two permanent meteorological stations: at a ridge site on the edge of the catchment (latitude: 43°03′N, longitude: 116°45′W, elevation: 2097 m), and at a sheltered site centrally located in the basin 350 m from the ridge site (known as the “snow pillow site”; elevation: 2061 m).

Details of the Reynolds Mountain East catchment and the instrumentation at the ridge and snow pillow sites are given by Slaughter et al. (2001). Marks and Winstral (2001) examined long-term data for these sites and found a difference of 30%–132% in annual precipitation at these two sites. They examined the difference in mass and energy fluxes to illustrate the importance of accounting for spatial distribution of model inputs in the simulation of snow deposition patterns. For this experiment, a third meteorological station was installed beneath a fir canopy adjacent to the snow pillow site.

Open site hourly averaged measurements of downwelling solar radiation, air temperature, wind speed, relative humidity, snow depth, and precipitation from both shielded and unshielded gauges are available at both ridge and snow pillow sites. Downwelling thermal radiation is additionally available at the snow pillow site, and diffuse radiation is measured with a shadowband radiometer at the ridge site. The climate station at the fir site measures subcanopy downwelling solar radiation, air temperature, relative humidity, wind speed, and snow depth beneath the fir canopy.

In addition to the automatic measurements made at the climate stations, snow pits were dug at the ridge (four pits) and fir (two pits) sites, where snowpack profiles of temperature, density, and snow grain size were measured. A thermistor array was installed at the fir site to provide continuous snow temperature profile measurements for the evaluation of SNOWCAN. Subcanopy upwelling and downwelling solar and thermal radiation measurements were taken over a 20-day period using two pyranometers and two pyrgeometers (one of each instrument was inverted to measure upwelling radiation). An array of nine upward-looking pyranometers was deployed at random locations beneath the fir canopy to capture the variability of solar radiation over five diurnal cycles. These radiometers were moved and releveled daily. A summary of the measurements available at each site is given in Table 1. Details of the climate station instrumentation are given by Hanson et al. (2001), the snow measurement instruments are described by Marks et al. (2001), and the precipitation gauges are discussed by Hanson (2001).

Difficulties were experienced with radiometer icing at the snow pillow site; therefore, solar radiation measured at the ridge site and simulated thermal radiation were used to drive the model at all sites. Simulation of thermal radiation was based on the method of Marks and Dozier (1979), where cloud cover was estimated from the difference between simulated and observed solar radiation (and interpolated at night). Underestimation of precipitation is known to occur at the windblown site; therefore an algorithm developed by Hanson et al. (2004) was used to estimate the actual precipitation for the ridge site from the shielded and unshielded gauge data. Other measurements at the ridge were used to drive the model for the ridge simulation. Measurements (except radiative measurements) from the snow pillow site were used to drive the model for the snow pillow and fir site simulations, which included precipitation measurements from the shielded gauge at the snow pillow site. Parameters from the snow pillow simulation were also used for the simulation of snow beneath the fir canopy.

A linear trend line for the scattergraph of fir site and snow pillow snow depths prior to significant melt on day 168 (excluding snow depths below 5 cm) showed that the fir site snow depth was approximately 53% of the snow depth at the snow pillow site during the accumulation period. Given the sheltered nature of the sites, the majority of the 47% difference could be explained by sublimation of canopy interception rather than by blowing snow processes. However, 47% is outside the range of measurements suggested by Pomeroy et al. (1998). For this evaluation, it was assumed that an arbitrary upper range value of 42% of precipitation was intercepted by the fir canopy and was lost to the atmosphere by sublimation. This assumption was tested by sensitivity analysis.

The method of Pierce and Running (1988) was used to determine the effective canopy leaf area index (LAI) from transmittance measured with a Decagon SF-80 Sunfleck Ceptometer. The effective LAI was measured as 4.0 m2 m−2, although retrieval of LAI with the Sunfleck Ceptometer assumes a spherical distribution of canopy elements. The effect of the canopy element distribution is discussed later. The spectral reflectance of the canopy elements was measured in a laboratory with a GER 3700 Spectroradiometer and was averaged to determine the canopy broadband single-scatter albedo. Images of the snow surface showed a maximum litter coverage of 7% at the end of the season, which resulted from merging snow layers; therefore litter deposition was assumed to be negligible for this study. For these simulations, the canopy emissivity was assumed to be 0.977.

4. Results

a. Open sites

Simulation of snow depth at the ridge site is shown in Fig. 2. Compared to measurements from the automatic depth sensor, simulated snow depth exceeds the observed depth from day 76 to the end of the season. However, measurement errors (both for precipitation and depth) associated with windy sites such as this are evident in Fig. 2, with large differences between measured precipitation (input to the model) and accumulation measured by the depth sensor. An example of this occurs on day 59, where more accumulation was observed on the ground than was simulated in the model. Accumulation (relative to pre-precipitation events) was also underestimated on days 121 and 145. In contrast, accumulation is overestimated in the model on day 104. These discrepancies between the measured and observed snow depth are reflected in the regression coefficient r2 of 0.88.

The most significant difference between the model and observations occurs on day 76. On this day, measured snow depth drops from 0.42 to 0.29 m in only 2 h, and is likely to be caused by an error in the sensor. It is not clear whether subsequent depth measurements are accurate, but it is from this day that snow depth is overestimated in the model when compared to these measurements. Compared to depth measured from snow pits, simulated snow depth is overestimated on day 122, but is consistent with depth measured from snow pits on days 140, 167, and 199.

A comparison between simulated and measured snow temperature, density, and grain size profiles is shown in Fig. 3. The temperature profiles simulated by the model follow the observed trends, although the simulated temperatures are warmer than the observations on days 122 and 167. Snow density profiles are simulated to within measurement error, except at the base of the snowpack, where the simulated snow is more dense than was measured. A peak in snow density on day 167 was simulated 61 cm above the base of the snowpack, and was observed at 65 cm. On day 199, this peak was simulated at 38 cm, compared to the observed peak at 35 cm above the snow–soil interface.

Simulated snow grain diameters agree with the available observed snow grain sizes to within measurement error on days 167 and 199 (Figs. 3i and 3l). A comparison with a more detailed snow grain size profile on day 140 (Fig. 3f) shows that the model simulates the larger grain sizes and surface grain sizes to within measurement error, but does not capture the small grain sizes observed 2–9 and 14–23 cm above the base of the snowpack.

Snow depth and mass simulated at the sheltered open site are shown in Fig. 4. The agreement with observations at this site is much better than at the windblown site as shown by regression coefficients of r2 = 0.98 for both depth and mass at the sheltered site. A precipitation event on day 29 led to the formation of the simulated snowpack four days before the observed snowpack. This may be why simulated snow mass exceeded observed mass in the accumulation period, until the melt event beginning on day 170, although the difference could also result from underestimation of measured snow mass from bridging of the snow pillow, as described by Beaumont (1965). At the end of the season, the rate of snowmelt is underestimated by the model, and the simulated snowpack had melted on day 225—seven days after the observed ground became snow free.

b. Fir site

At the fir site, snow depth simulated beneath the forest canopy shows good agreement with measurements (r2 = 0.96), as shown in Fig. 5. As for the snow pillow site, the rate of snowmelt is underestimated by the model. At the fir site the ground beneath the canopy became snow free on day 213 in the simulation—five days later than the observations, although simulated snow ablation was delayed by precipitation events on days 209 and 212.

A comparison between simulated and measured snow density profiles, as shown in Figs. 6b and 6e, demonstrates that the model simulated the density to within measurement error. The density peak on day 139 measured at 15 cm was simulated at 18 cm. On day 167, the peak–trough–peak in density 15–35 cm above the snow base was also captured by the model. The snow grain size profile on day 139 was simulated within the measurement accuracy, but the grain size of the melt–freeze layer at 26–30 cm was overestimated by the model on day 167.

Snow temperature profiles shown in Figs. 6a and 6d are generally warmer than observations, and liquid water is simulated in the lower 25 cm of the snowpack on day 167. This is also shown in the comparison with the continuous measurements given in Fig. 7. The simulated trends follow the observations at all depths shown in Fig. 7 until day 164, when the simulated snowpack becomes ripe but the observations indicate that it is not.

Simulation of the reduction of solar radiation beneath the canopy is shown in Fig. 8. The forcing data are included in this graph to demonstrate the magnitude of shielding by the canopy. Simulated subcanopy downwelling solar radiation is of the correct order of magnitude compared to the measurement from one radiometer in this period, although the measurements are more variable on sunny days. Multiple reflections of solar radiation at this site contribute an average daily total of 39.6 W m−2 to the downwelling solar radiation, and are more significant in the ablation season.

Figure 9 allows the differences between simulated and measured subcanopy solar radiation to be examined in greater detail, by focusing on a 5-day period when measurements from an array of radiometers were also available. Figure 9a compares simulated downwelling solar radiation with the average measurements from the array, whereas Fig. 9b shows the upwelling radiation compared to the measurement from one radiometer. For the downwelling radiation, the agreement between the simulated radiation and array average is r2 = 0.72 for the sunny days (199 and 200) and is r2 = 0.91 for the diffuse days. The dip in measured upwelling radiation on the sunny days was caused by the shadow of a tree trunk falling across the radiometer and is not reproduced by the model.

Simulated subcanopy thermal radiation is shown in Fig. 10. The forcing data are included to demonstrate the enhancement of thermal radiation by the canopy. Simulated radiation follows the trend in measurements closely (r2 = 0.95 and 0.87 for downwelling and upwelling, respectively). The upwelling thermal radiation is limited by the melting temperature of the snow on days 197–200. Simulated thermal radiation exceeds the measurements for both up- and downwelling radiation by approximately 10 W m−2. The systematic overestimation in thermal radiation is not caused by the parameterization of the emissivity, as this phenomenon occurs with snow and canopy emissivity equal to 100%. It is more likely to be due to the dome effect of the pyranometers, a modification of the radiation balance between the sensor and target that results in an underestimation of thermal irradiance by more than 10 W m−2 (Ji and Tsay 2000).

5. Discussion

This study examines detailed snow characteristics at open sites and beneath a fir canopy simulated by SNOWCAN. Snow properties at two open sites were simulated: a windblown ridge site, where temperature, density, and grain size profiles were measured; and a sheltered open site, where continuous measurements of snow mass were available. Snow mass and depth at the sheltered site showed good agreement with observations but the agreement with automatic depth measurements was less good at the windblown site. However, simulated snow depth at the ridge site was consistent with depth measured at three of the four snow pits.

The accuracy of snow depth simulation by SNOWCAN is highly dependent on the quality of precipitation data. Precipitation is difficult to measure, particularly in windy environments. A comparison with automatic depth measurements is also complicated by blowing snow processes, which are not simulated by SNOWCAN. In its present form, SNOWCAN cannot account for spatial variability in snow depth.

Where simulated snow depth is consistent with the snow pit measurement, the temperature profile follows the trend of the observations. Where the model does not simulate snow depth accurately, on day 122, the simulated temperature profile is more steep than the observed profile. This could be caused by greater energy input into the simulated snowpack, or occur because there is more mass to cool in the model and therefore less cooling in each layer. Without continuous temperature measurements or complete energy balance measurements, it is not possible to determine the cause of the temperature profile differences. However, for a site with high wind speeds such as this ridge site, errors in the turbulent fluxes could be significant. Good quality measurements of turbulent fluxes over snow are needed as part of a suite of measurements to test all components of the mass and energy balance.

The density and grain size profiles are less sensitive to snow depth accuracy than the temperature profile. Density profiles agree with measurements to within measurement error above the lower ∼10 cm of the snowpack base. An overestimation of the density at the base suggests excessive pressure compaction relative to metamorphic densification. It is crucial to simulate the magnitude of densification accurately, as with a poor densification parameterization, the temperature profiles are not accurate (simulations not shown), and the date of complete snow ablation may vary by several days.

At the sheltered open site, the simulation of snow mass and depth is more straightforward, as shown by the good agreement with observations. Snow mass exceeds the measurements in the accumulation season. The difference occurs early in the season and throughout the accumulation period: the snowpack is formed earlier in the simulation than was observed. Possible causes of this are the temperature of the ground as the snow is deposited and the classification of precipitation, that is, the snow/rain partition in the model. Alternatively, bridging of the snow pillow may have occurred, which would lead to an underestimation of the measured snow mass.

Tribbeck et al. (2004) showed that simulation of trends in near-surface soil temperature was possible with a constant temperature at the lower boundary However, at the beginning of the season, errors of the order of 1°C may allow snow to settle on the ground instead of melting, or melt to occur in the model where snow deposition was observed. In this regard, SNOWCAN would benefit from an improved treatment of the soil, such as input of measured soil temperatures (assuming these are available) and consideration of water transport within the soil.

Evaluation of SNOWCAN at the open sites shows the accuracy and limitations of the snow component in the model. Application of the input data and parameterization at the sheltered open site to the fir site simulation enables the evaluation of the canopy component and the interaction between the canopy and snow in the model. In common with the snow pillow site, simulated and observed snow depths show good agreement, although, again, snowmelt in the model occurs more slowly than the observed rate.

An examination of simulated internal snow variables reveals differences between the open and fir sites. Density profiles at the fir site show much more variation than at the ridge site and the sheltered site (not shown). Pressure densification is reduced in the shallower snowpack, so densification by metamorphic processes plays a greater role than in the deeper snowpacks at the open sites.

At the fir site, trends in the simulated density profiles can also be observed in the grain size profiles, for example, large increases in density at similar depths to large increases in the grain size. Both density and grain size profiles contain information on the thermal history of the snowpack as modeled metamorphism densification is dependent on the temperature, and grain growth in dry snow through hand-to-hand vapor diffusion is governed by the temperature profile. Both these processes occur more rapidly in wet snow.

The temperature profile in the simulated fir site snowpack on day 139 was warmer than the measurements, and the simulated snowpack was deeper than observed. This is similar to the ridge site on day 122 (the model overestimated the mass of snow and the heat stored in the snowpack). On day 167, however, ripe snow was simulated at the base of the fir snowpack, but not in the ridge snowpack. A ripe snowpack base was also simulated at the sheltered open site (not shown), which indicates that the discrepancy between the simulation and observations is driven by the snow processes rather than the representation of the vegetation canopy.

The continuous snow temperature profiles at the fir site indicate that liquid water does not drain from the snow and/or refreeze following the melt event on day 164. Gravitational liquid water drainage is assumed in SNOWCAN and the liquid water flux is related nonlinearly to the density through the hydraulic permeability. Slow drainage at the fir and sheltered sites could be a result of the low density simulated at the base of the snow. Low density at the base of the pack probably results from the assumption that no vapor transport occurs across the snow–soil interface; thus no vapor is available to replace that lost from the base of the snow and transported up through the snowpack by hand-to-hand diffusion. Alternatively, liquid water may be drained more efficiently in the field by processes other than gravitational drainage, such as through rapid drainage channels, which are not represented in SNOWCAN.

Slow drainage of liquid water could also contribute to the underestimation of melt rate at the end of the season, as mass is retained within the snowpack through melt–freeze cycles. Late melt-out is also simulated at other sites and with other models (e.g., Tribbeck et al. 2004; Hardy et al. 1997; Lehning et al. 1999; Link and Marks 1999). The reason for this common problem is not clear; advection of heat from snow-free patches would enhance melt rates, as described by Shook and Gray (1997), but the melt rate is underestimated in snowpacks of 1 m or deeper (where the snow would not be patchy). The melt rate and liquid water drainage of snow requires further investigation.

Subcanopy solar radiation is highly variable on sunny conditions because sunflecks beneath the canopy change with sun angle, cloud, and canopy movement. Some of the measured variations are reduced by averaging the measurement from an array of radiometers. At this site, the array-averaged measurement still shows high variation, and nine radiometers were insufficient to provide a smooth profile ideal for comparison with the modeled solar radiation on sunny days. On diffuse days, the simulated solar radiation is in good agreement with the measurement from a single radiometer, as well as with the array-averaged measurement.

Thermal radiation incident on the snow beneath the fir canopy was observed to be up to 100 W m−2 greater than that in the open. This enhancement of thermal radiation by the vegetation canopy was also simulated by SNOWCAN. Given the close match between the observed and simulated downwelling thermal radiation, the assumption that the canopy is at the air temperature is appropriate for this site, but should be tested for a canopy with lower LAI, where the thermal radiation is more dependent on the canopy density.

The main difference between simulated and observed thermal radiation occurred in the upwelling radiation in the period where the simulated snowpack was melting but the observed snowpack did not appear to be (from the temperature measurements). Neither the thermal nor solar radiation components demonstrate that the early melt of the simulated snowpack is caused by excessive radiative energy input. Given that early snowmelt was simulated beneath the fir canopy and at the sheltered open site, the source of error is common to both. This could lie in forcing data measurement errors, or in the representation of liquid water processes in the model.

Although in this study the sources of error appear to lie within the snow model, errors in the canopy model could become more important in the application of SNOWCAN to other canopy sites, and may become apparent if data were collected to assess the complete mass and energy budget. One model aspect that could be examined is the algorithm for reduction of wind speed beneath the canopy. Some measurements of subcanopy wind speed were available in the season. Given that the maximum wind speed observed was 2.93 m s−1, the turbulent transfer is likely to be low. Tribbeck (2002) simulated the ablation period at this site with the wind speed algorithm and separately with the observed subcanopy wind speed. Use of the algorithm as an alternative to the measurements had negligible impact on the simulation of the snow cover. However, this assumption could be tested in greater detail if measurements of sensible and latent heat were available.

Another simplifying assumption applied here is in the treatment of canopy interception and sublimation. Throughout the field experiment described, the assumption that all intercepted snow sublimates is appropriate as the snow did not stay on the trees for more than a few days, wind speeds were low, and the canopy branches did not appear to be overloaded. At the end of the season, interception melt and drip occurred on at least one day, and may have contributed to faster snowmelt because of the heat transfer from the water (similar to the effect of rain).

The parameterization of interception loss was tested by sensitivity analysis. Although a comparison between fir and snow pillow snow depths during the accumulation period suggested that the parameter may be as high as 47%, a value of 42% was chosen, which is in the upper range of measurements detailed in the literature (Pomeroy et al. 1998). With a higher interception loss of 45%, the agreement between measured and simulated depth at the fir site improves (r2 = 0.97), but the simulated density of the snowpack is too low. With a lower interception loss of 40%, the agreement between simulated and measured depth decreases. Improved simulations may be possible by adjusting the compaction parameterization, but this in turn also affects the snow depth. Unfortunately, continuous measurements of snow water equivalent are not available beneath the fir canopy (which would allow us to examine this assumption in greater detail). Ideally, a full mass and energy budget of the canopy is required to simulate interception processes. This is beyond the scope of this study, which concentrates on the radiative interaction between the canopy and snow.

In the calculation of the radiative transfer, a spherical distribution of elements was assumed. For this study, a Decagon Sunfleck Ceptometer was used to determine the model leaf area index parameter. Retrieval of the leaf area index with this instrument is based on an assumption of spherical distribution of canopy elements; hence the measurement relates to the radiation interception efficiency (and is an effective leaf area index), but not biomass of the canopy (Decagon Devices 1991). The assumption of a spherical distribution of canopy elements is appropriate here because the effective leaf area index has been used. Even so, the sensitivity of the model to the assumption of spherical distribution can be tested.

In a spherical distribution, the canopy elements have an equal probability of alignment in any direction. However, the true distribution could lie between all elements horizontally aligned (planophile) and all elements vertically aligned (erectophile). The directional optical depths for a planophile and an erectophile canopy are defined in terms of the leaf area index as τ = cos θs · LAI and τ = (2/π)sin θs · LAI, respectively, where θs is the solar zenith angle (Roujean 1996). These relations cannot be used in SNOWCAN directly because the thermal radiation is also dependent on the third exponential distribution, which cannot be calculated at night when the cosine of the solar zenith angle is negative. However, simple ratios between the LAI and optical depth can be derived from these equations for planophile and erectophile canopies, by integration of these functions over the range of solar zenith angles. These have been used in place of the simple ratio arising from the assumption of a spherical canopy distribution to examine the sensitivity of SNOWCAN to the spherical distribution assumption.

For this site, the mean optical depth for the season is 0.41 · LAI for a planophile canopy and 0.55 · LAI for an erectophile canopy. With these parameterizations, snow beneath a planophile canopy melts 50 h later than beneath the spherical canopy, while the snow beneath an erectophile canopy melts 30 h earlier. The agreement with measured snow depth improves for the erectophile canopy over the spherically distributed canopy, and the simulated melt date is closer to the observed date. However, because of the shape of the third exponential integral the model will be more sensitive to the canopy distribution assumption at low values of the leaf area index.

This study demonstrates the potential of SNOWCAN to simulate the evolution of snow beneath a forest canopy and investigate the effects of a forest cover on snow. A more rigorous treatment and evaluation of turbulent transfer and snow interception is required to extend the capabilities and applications of SNOWCAN, but is beyond the scope of this study. The radiative components of the energy balance and snowpack properties have been evaluated here. Errors have been found with the onset of melt, but it is not likely that these result from the radiation components of the model, which show good agreement with measurements. The canopy assumptions appear to be valid at this site but need to be tested at other sites with lower canopy densities.

Although SNOWCAN is not adapted to windy environments, the simulated snow depth was comparable to the depths of three snow pits at a windblown open site, which allowed comparisons of internal snow profiles. These, and measurements beneath the fir canopy demonstrated that the representation of the snow physics was reasonable prior to significant snowmelt. With the onset of melt, however, too much liquid water was retained in the simulated snowpack. Either the rate of refreeze or the rate of drainage, or both, were not simulated accurately and require further investigation.

With improvements to the processes indicated, this model can be used to derive simpler representations of snowpack and forest interactions that are still physically based, and are therefore applicable to any site. Simplified representations in turn can be incorporated into models over a range of scales, including general circulation models, to improve their accuracy.

Acknowledgments

This work was funded in part by NERC Studentship GT 04/98/TS/247 and a CASE award from the British Antarctic Survey. Many thanks are due to Danny Marks, Adam Winstral, and the USDA/ARS Northwest Watershed Research Center for collaboration in the field and data analysis. We also thank the chief editor, Dr. Kustas, and three anonymous reviewers who helped to improve this paper.

REFERENCES

  • Adams, R. S., , Spittlehouse D. L. , , and Winkler R. D. , 1996: The effect of a canopy on the snowmelt energy balance. Proc. 64th Annual Meeting Western Snow Conf., Bend, OR, Western Snow Conference, 171–174.

  • Beaumont, R. T., 1965: Mt. Hood pressure pillow snow gauge. J. Appl. Meteor., 4 , 626631.

  • Betts, R. A., 2000: Offset of the potential carbon sink from boreal forestation by decreases in surface albedo. Nature, 408 , 187190.

  • Chen, J., , Rich P. W. , , Gower S. T. , , Norman J. M. , , and Plummer S. , 1997: Leaf area index of boreal forests: Theory, techniques and measurements. J. Geophys. Res., 102 , 2942929444.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Decagon Devices, 1991: Sunfleck PAR Ceptometer operator’s manual. Decagon Devices, Pullman, WA, 77 pp.

  • Golding, D. L., , and Swanson R. H. , 1986: Snow distribution patterns in clearings and adjacent forest. Water Resour. Res., 22 , 19311940.

  • Hanson, C. L., 2001: Long-term precipitation database, Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28312834.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hanson, C. L., , Marks D. , , and Van Vactor S. S. , 2001: Long-term climate database, Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28392841.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hanson, C. L., , Pierson F. B. , , and Johnson G. L. , 2004: Dual-gauge system for measuring precipitation: Historical development and use. J. Hydrol. Eng., 9 , 350359.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hardy, J. P., , and Hansen-Bristow K. J. , 1990: Temporal accumulation and ablation patterns of the seasonal snowpack in forests representing varying stages of growth. Proc. 58th Western Snow Conf., Sacramento, CA, Western Snow Conference, 23–34.

  • Hardy, J. P., , Davis R. E. , , Jordan R. , , Li X. , , Woodcock C. , , Ni W. , , and McKenzie J. C. , 1997: Snow ablation modeling at the stand scale in a boreal jack pine forest. J. Geophys. Res., 102 , 2939729405.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hardy, J. P., , Melloh R. , , Robinson P. , , and Jordan R. , 2000: Incorporating effects of forest litter in a snow process model. Hydrol. Processes, 14 , 32273237.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Henderson-Sellers, A., , and Wilson M. F. , 1983: Surface albedo data for climatic modeling. Rev. Geophys. Space Phys., 21 , 17431778.

  • Ji, Q., , and Tsay S. , 2000: On the dome effect of Eppley pyrgeometers and pyranometers. Geophys. Res. Lett., 27 , 971974.

  • Jordan, R., 1991: A one-dimensional temperature model for a snow cover: Technical documentation for SNTHERM. CRREL Tech. Rep. 91-16, 64 pp.

  • Jordan, R., 1992: Estimating turbulent transfer functions for use in energy balance models. CRREL Tech. Rep. 1107, 17 pp.

  • Koivusalo, H., , and Kokkonen T. , 2002: Snow processes in a forest clearing and in a coniferous forest. J. Hydrol., 262 , 145164.

  • Lehning, M., , Bartelt P. , , Brown B. , , Russi T. , , Stöckli U. , , and Zimmerli M. , 1999: SNOWPACK model calculations for avalanche warning based upon a new network of weather and snow stations. Cold Reg. Sci. Technol., 30 , 145157.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Link, T. E., , and Marks D. , 1999: Point simulation of seasonal snow cover dynamics beneath boreal forest canopies. J. Geophys. Res., 104 , 2784127857.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks, D., 1988: Climate, energy exchange, and snowmelt in Emerald Lake Watershed, Sierra Nevada. Ph.D. thesis, University of California, Santa Barbara, 158 pp.

  • Marks, D., , and Dozier J. , 1979: A clear-sky longwave radiation model for remote alpine areas. Arch. Meteor. Geophys. Bioklimatol., 27B , 159187.

    • Search Google Scholar
    • Export Citation
  • Marks, D., , and Winstral A. , 2001: Comparison of snow deposition, the snow cover energy balance, and snowmelt at two sites in a semiarid mountain basin. J. Hydrometeor., 2 , 213227.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks, D., , Cooley K. R. , , Robertson D. C. , , and Winstral A. , 2001: Long-term snow database, Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28352838.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marshall, S. E., , and Warren S. G. , 1987: Parameterization of snow albedo for climate models. Large Scale Effects of Seasonal Snow Cover, B. E. Goodison, R. G. Barry, and J. Dozier, Eds., IAHS Publication 166, 43–50.

    • Search Google Scholar
    • Export Citation
  • Marshall, S., , Roads J. O. , , and Glatzmaier G. , 1994: Snow hydrology in a general circulation model. J. Climate, 7 , 12511269.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pearson, D., , Daamen C. C. , , Gurney R. J. , , and Simmonds L. P. , 1999: Combined modelling of shortwave and thermal radiation for one-dimensional SVATs. Hydrol. Earth Syst. Sci., 3 , 1530.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pierce, L. L., , and Running S. W. , 1988: Rapid estimation of coniferous forest leaf area index using a portable integrating radiometer. Ecology, 69 , 17621767.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pomeroy, J. W., , Parviainen J. , , Hedstrom N. , , and Gray D. M. , 1998: Coupled modelling of forest snow interception and sublimation. Hydrol. Processes, 12 , 23172337.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Price, A. G., , and Dunne T. , 1976: Energy balance computations of snowmelt in a sub-arctic area. Water Resour. Res., 12 , 686694.

  • Roujean, J. L., 1996: A tractable physical model of shortwave radiation interception by vegetative canopies. J. Geophys. Res., 101 , 95239532.

  • Rouse, W. R., 1984: Microclimate at arctic tree line. 1. Radiation balance of tundra and forest. Water Resour. Res., 20 , 5766.

  • Shook, K., , and Gray D. M. , 1997: Snowmelt resulting from advection. Hydrol. Processes, 11 , 17251736.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Slaughter, C. W., , Marks D. , , Flerchinger G. N. , , Van Vactor S. S. , , and Burgess M. , 2001: Thirty-five years of research data collection at the Reynolds Creek Experimental Watershed, Idaho, United States. Water Resour. Res., 37 , 28192823.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thomas, G., , and Rowntree P. R. , 1992: The boreal forests and climate. Quart. J. Roy. Meteor. Soc., 118 , 469497.

  • Tribbeck, M. J., 2002: Modelling the effect of vegetation on the seasonal snow cover. Ph.D. thesis, University of Reading, 240 pp.

  • Tribbeck, M. J., , Gurney R. J. , , Morris E. M. , , and Pearson D. W. C. , 2004: A new Snow-SVAT to simulate the accumulation and ablation of seasonal snowcover beneath a forest canopy. J. Glaciol., 50 , 171182.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Viterbo, P., , and Betts A. K. , 1999: The forecast impact of changes to the albedo of the boreal forests in the presence of snow. J. Geophys. Res., 104 , 2780327810.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wiscombe, W. J., , and Warren S. G. , 1980: A model for the spectral albedo of snow. I. Pure snow. J. Atmos. Sci., 37 , 27122733.

    • Crossref
    • Search Google Scholar
    • Export Citation
Fig. 1.
Fig. 1.

Flowchart description of SNOWCAN.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 2.
Fig. 2.

Snow depth at an open, windblown site.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 3.
Fig. 3.

Snow temperature, density, and grain size profiles at an open, windblown site on (a)–(c) day 122, (d)–(f) day 140, (g)–(i) day 167, and (j)–(l) day 199. Two sets of measurements were made on day 122, shown separately by different shading.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 4.
Fig. 4.

Depth and mass of snow at an open, sheltered site.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 5.
Fig. 5.

Comparison between observed and simulated snow depth beneath a fir canopy.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 6.
Fig. 6.

Evaluation of snow temperature and density profiles beneath a fir canopy on (a)–(c) day 139 and (d)–(f) day 167.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 7.
Fig. 7.

Continuous simulated and measured snow temperature at (a) 30, (b) 20, and (c) 10 cm above the snow–soil interface and (d) at the interface.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 8.
Fig. 8.

Reduction of solar radiative flux beneath a fir canopy.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 9.
Fig. 9.

Subcanopy solar radiation in a 5-day period: (a) simulated downwelling radiation compared to array-averaged measurements and (b) simulated upwelling radiation compared to a single radiometer.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Fig. 10.
Fig. 10.

Subcanopy longwave radiation (a) downwelling and (b) upwelling.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM528.1

Table 1.

Summary of field campaign measurements.

Table 1.
Save