High‐resolution simulations of the night‐time stable boundary layer over snow

Abstract

The nocturnal stable boundary layer (SBL) over snow was studied via 1-D clear-sky simulations at 67 o N. Three wind regimes could be identified. A windy SBL was relatively well-mixed, turbulent cooling rates being moderate and long-wave (LW) cooling rates weak as gradients of temperature and moisture (T,q) remained small. In calm and near-calm cases the surface temperature T s dropped rapidly in the evening, leading to a strong shallow inversion dominated by strong LW cooling. During weak winds (geostrophic wind V g ∼ 3 m s −1 ), T s decreased fairly rapidly in the evening with T 2m closely coupled. The nocturnal inversion was moderately strong with strong turbulent cooling in the surface layer. The T,q-profiles induced moderate LW cooling at mid-inversion levels but LW heating in the lowest metres, which resisted the strong turbulent cooling. Thus the heating/cooling rates were fairly sensitive to the wind, especially near the surface.The snow surface was coldest after a calm night and warmer at windy sunrises, whereas the coldest air temperatures were obtained during weak winds, provided that the model's stability functions were of the 'short tails' type. A maximum in the downward sensible heat flux was associated with this. When the model's multilayer snow scheme was replaced by a coarse scheme, the nocturnal T s and T 2m stayed too warm. The use of old snow properties instead of typical snow produced relatively warm nocturnal temperatures whereas fresh snow led to quite cold temperatures. These aspects may have relevance to the SBL warm biases of large-scale models.

Clear‐sky stable boundary layers with low winds over snow‐covered surfaces. Part 1: WRF model evaluation

“…Also, when we compare the radiative flux divergence with the turbulent flux divergence in Eq. for set D, we find for Cabauw close to the surface a turbulent flux divergence varying from −1 K h −1 at the first hour to −0.24 K h −1 at the end of the night, while the radiative flux divergence has a more continuous value of ∼−0.15 K h −1 , indicating the importance of radiative processes at these low wind speeds (Savijärvi, , ). The advection is rather small close to the surface, but becomes relatively more important higher up.…”

“…This appeared to be very important for wind speed. Due to the fact that this was a low wind‐speed case, turbulence is relatively weak and thus the turbulent flux divergence becomes relatively small and the other terms in Eqs become relatively more important (Savijärvi, , ). Also, when we compare the radiative flux divergence with the turbulent flux divergence in Eq.…”

“…in operational models) can also hamper proper SBL modelling (e.g. Steeneveld et al, 2006;Tardif, 2007;Byrkjedal et al, 2008;Svensson and Holtslag, 2009;Svensson and Karlsson, 2011;Savijärvi, 2013). Especially over snow-covered polar surfaces, where atmospheric conditions can become very stable, modelling the SBL is challenging.…”

Current challenges in understanding and forecasting stable boundary layers over land and ice

“…Since these variations are large close to the surface, in particular for calm conditions, one may expect substantial radiation divergence near the surface. Indeed, numerous modeling studies reported such a divergence (Ha and Mahrt, 2003;Savijärvi, 2013). Field observations by Hoch et al (2007) and Steeneveld et al (2010) (Figure 1) reported radiation divergence values of several K/h in favorable conditions, particularly during sunset.…”

Clear‐sky stable boundary layers with low winds over snow‐covered surfaces. Part 2: Process sensitivity

“…for fresh snow with a density of 100 kg m −3 , a λ snow of 0.128 W m −1 K −1 is utilized (where we used 0.021 W m −1 K −1 for Cabauw observations), while for snow with density 200 kg m −3 , a λ snow of 0.215 W m −1 K −1 is modelled (we used 0.084 W m −1 K −1 for Sodankylä observations). Savijärvi (2013) also reported on the strong impact of the selected density and conductivity on the (near-)surface temperatures, especially in stable conditions. Figure 4 confirms that reducing λ snow brings the model results closer to the observations for Cabauw and Sodankylä, though for the latter G is only slightly overestimated.…”