Snow albedo

The albedo of a snow-covered surface is influenced by many factors, including the depth and grain structure of the snow, contaminants in the snow, the albedo of the underlying surface, heterogeneities in the snow cover, and masking by vegetation (Sections 2.5, 3.4). The simplest GCM snow models neglect these influences and assign a fixed albedo to any gridbox with snow cover. Snow albedos also vary greatly with the wavelength of incident radiation; although they split the solar spectrum into bands for calculating radiative transfers in the atmosphere, GCMs often use a single value of surface albedo for all wavelengths.

BASE, SiB, and the original version of MOSES represent snow ageing by making the albedo a function of temperature, decreasing as the temperature approaches the melting point. This approach gives an unrealistic increase in albedo when melting snow refreezes. A better representation is given by parameterizing the albedo as a function of the age of the snow surface. CLASS, ISBA, and MPI use exponential or linear relationships to increment snow albedos according to

as(t + At) = [as(t) - «mini exp I - —I + «mm or as(t + At) = max

Ta where amin is a lower limit on the snow albedo and the empirical time constant Ta is shorter for melting snow than cold snow. Values have to be assigned for the albedo of fresh snow and the depth of snowfall required to refresh the surface albedo. Maximum and minimum values of albedo are specified separately for visible and near-infrared bands in CLASS. In GISS83, the snow albedo is given by as = amin + («max - «min) exp(-a/5), for snow surface age a (in days) updated according to

V dc where Ads is the snowfall in a timestep and dc is the depth required to refresh the albedo.

Snow age (days)

Figure 4.12. Albedo decay with age for melting snow from parameterizations used by CLASS (—), GISS83 (■ ■ ■) and BATS (---).

Marshall (1989) used results from a spectral snow albedo model (Warren and Wiscombe, 1980; Wiscombe and Warren, 1980) to develop a physically based parameterization for implementation in the NCAR GCM (Marshall and Oglesby, 1994). This parameterization has also been adopted as an option in MOSES (Essery et al., 2003a). Given snow depth, grain size, soot content, and zenith angle, separate diffuse and direct beam albedos are calculated for visible and near-infrared wavelength bands. Grain growth rates are taken from measurements in cold and melting snow. BATS and CLM use a similar parameterization with a snow age parameter in place of the grain size. Figure 4.12 shows the albedo decay for melting snow in BATS, CLASS, and GISS83.

Many models parameterize gridbox-average albedos as functions of snow depth, since the influences of heterogeneity, vegetation masking, and absorption of radiation by the underlying surface are likely to be more significant for shallow snow. The albedo is typically taken to be a weighted average a = fs^s + (1 - fs)a0, where as is the albedo of deep, homogeneous snow and a0 is the albedo of the snow-free surface. Functions used for the weighting have included

HS + a2 zo used by BATS, CLM, and ISBA, fs = 1 - e-a2HS

used by GISS83 and MOSES, and fs = min[a2HS, 1]

Snow depth (cm)

Figure 4.13. Snow-cover fraction as a function of average snow depth from param-eterizations used by CLASS (—), MOSES (■ ■ ■) and ISBA with z0 = 10-3 m (---)

used by CLASS and SiB, where HS is the snow depth, z0 is the surface roughness length, and a2 is a parameter; the results are shown in Figure 4.13. Separate masking functions may be used for the vegetated and bare-ground fractions of a gridbox, and maximum snow-covered albedos may be assigned according to vegetation type. ISBA and a new scheme for the ECHAM4 GCM (Roesch et al. ,2001) additionally allow for subgrid distributions of snow with elevation by parameterizing the snow-cover fraction as a function of the standard deviation of surface heights within a gridbox.

Little theoretical or observational justification has been offered as yet for the parameterizations of snow-covered area and vegetation masking currently used in GCMs (Essery and Pomeroy, 2004), but attempts have been made to relate snow mass and fractional area to land-cover classes (e.g. Donald et al., 1995; Liston, 2004). In an off-line test of BATS, Yang et al. (1997) found that an alternative function

gave a better match to observed snow-cover fractions for short vegetation.

4.4.4 Snow energy and mass balances

The components of snow surface energy and mass balances are discussed in Section 3.2. GCMs often neglect heat advected to the surface by precipitation and none, as yet, represents blowing snow. Surface roughness is reduced as snow depth increases or is calculated as an effective average for snow-covered and

Snow depth (cm)

Figure 4.13. Snow-cover fraction as a function of average snow depth from param-eterizations used by CLASS (—), MOSES (■ ■ ■) and ISBA with z0 = 10-3 m (---)

snow-free fractions of a gridbox, and snow-covered fractions are treated as saturated for the calculation of latent heat fluxes. The formulation of aerodynamic resistances was found to have an important influence on snow surface energy balances for a high-latitude environment in the PILPS 2e model intercomparison (Bowling et al., 2003).

The energy balance of a snowpack is determined by the net surface energy flux, heat fluxes between the snow and the underlying ground, and phase changes within the snow (melting and freezing). For numerical calculations of heat fluxes and temperature changes, a snowpack and the underlying ground have to be divided into layers. The conducted heat flux between two layers is parameterized as being proportional to the difference in their temperatures and a weighted average of their thermal conductivities. GCMs typically use between 2 and 6 layers to represent the top few meters of soil but often represent a snowpack as a single layer or combine it with the surface soil layer. With increases in computational power, the use of multi-layer snow models such as GISS94 and CLM, which allow for 3 or 5 snow layers respectively, is likely to become more common.

When the calculated snow surface temperature reaches 0 °C, subsequent net energy input from radiation, turbulent heat fluxes, or ground heat fluxes (Section 3.3) is used to melt snow. In ISBA, an empirical approach is introduced to partition energy between melting snow and the warming of any overlying vegetation canopy, allowing the surface temperature to rise above 0 °C even while there is snow on the ground. Models differ in whether they calculate a single energy balance for the composite surface or separate energy balances for the snow-covered and snow-free parts of a partially snow-covered gridbox; composite energy balance calculations can melt snow too early (Liston, 2004; Essery et al., 2005), whereas separate energy balances may lead to snow melting too late (Slater et al., 2001).

Several off-line studies (using observed meteorological data to drive a surface model rather than coupling it to a GCM) have found snow models to be very sensitive to variations in the prescribed downward longwave radiation (Yang et al., 1997; Slater et al. ,1998; Schlosser et al., 2000). Considering uncertainties in GCM simulations of longwave radiation at high latitudes, Slater et al. (1998) argued that the use of complex snow models in GCMs may not yet be justified.

4.4.5 Heterogeneous snow cover

As can be seen from Figure 4.9, a typical GCM land gridbox spans a large area and may include regions with large differences in elevation, aspect, and vegetation cover. Many "land" gridboxes will actually contain significant fractions of inland or coastal water. GCMs, however, generally assume that land surface properties are homogeneous within each gridbox or can be characterized by effective parameters. Snow cover is frequently heterogeneous on length scales too small to be resolved by a GCM grid, introducing marked inhomogeneities in land surface properties such as albedo, roughness, and moisture availability. Most models diagnose a fractional snow cover, as discussed above, and use it to calculate effective gridbox parameters.

First-order turbulence closure schemes relate surface fluxes of momentum, heat, and moisture to windspeed, temperature, and humidity gradients between the surface and an atmospheric reference level, typically 10-30 m above the surface in GCMs, using transfer coefficients that depend on atmospheric stability and the roughness of the surface (Section 3.3). Due to the non-linearity of relationships between local fluxes and local gradients, gridbox-average fluxes are not simply related to gridbox-average gradients over heterogeneous surfaces, and the specification of effective parameters is not straightforward (Mahrt, 1996). The average sensible heat flux over a surface with heterogeneous snow cover can be dominated by the contribution from a small fraction of warm, dry snow-free land, giving an upward average flux counter to a downward average temperature gradient (Essery, 1997b). It has been suggested that the problem of calculating gridbox-average surface fluxes can be addressed by gathering distinct surface types within a gridbox into homogeneous "tiles" and calculating fluxes separately over each tile (Avissar and Pielke, 1989; Liston, 1995; Essery et al., 2003a); this approach is adopted for snow-covered and snow-free fractions of gridboxes in CLASS. The TESSEL surface scheme (van den Hurk et al., 2000) allows exposed snow, snow beneath tall vegetation, and snow-free exposed fractions to coexist within a gridbox.

Snow cover will often be heterogeneous for mid-latitude gridboxes covering areas with significant subgrid variations in surface elevation. To represent this in a GCM, Walland and Simmonds (1996) developed a model that divides each gridbox into cold snow, melting snow, and snow-free fractions above and below a diagnosed freezing level. Liston et al. (1996) found that nesting a snow model with a 5 km resolution within a regional atmospheric model with a 50 km resolution significantly improved the simulation of snow cover over the Rocky Mountains in Colorado. Fortunately, such high resolution may not be necessary; comparing simulations of snowmelt over a mountainous 1° x 1° region in western Montana, Arola and Lettenmaier (1996) found that a distributed model gave very similar results whether used on a high-resolution grid or with just ten elevation bands.

Renewable Energy 101

Renewable Energy 101

Renewable energy is energy that is generated from sunlight, rain, tides, geothermal heat and wind. These sources are naturally and constantly replenished, which is why they are deemed as renewable. The usage of renewable energy sources is very important when considering the sustainability of the existing energy usage of the world. While there is currently an abundance of non-renewable energy sources, such as nuclear fuels, these energy sources are depleting. In addition to being a non-renewable supply, the non-renewable energy sources release emissions into the air, which has an adverse effect on the environment.

Get My Free Ebook

Post a comment