The values of the snow and soil thermal conductivity, k_snow and k_soil, strongly impact the thermal regime of the ground in the Arctic, but very few data are available to test model predictions for these variables. We have monitored k_snow and k_soil using heated needle probes at Bylot Island in the Canadian High Arctic (73° N, 80° W) between July 2013 and July 2015. Few k_snow data were obtained during the 2013–2014 winter, because little snow was present. During the 2014–2015 winter k_snow monitoring at 2, 12 and 22 cm heights and field observations show that a depth hoar layer with k_snow around 0.02 W m⁻¹ K⁻¹ rapidly formed. At 12 and 22 cm, wind slabs with k_snow around 0.2 to 0.3 W m⁻¹ K⁻¹ formed. The monitoring of k_soil at 10 cm depth shows that in thawed soil k_soil was around 0.7 W m⁻¹ K⁻¹, while in frozen soil it was around 1.9 W m⁻¹ K⁻¹. The transition between both values took place within a few days, with faster thawing than freezing and a hysteresis effect evidenced in the thermal conductivity–liquid water content relationship. The fast transitions suggest that the use of a bimodal distribution of k_soil for modelling may be an interesting option that deserves further testing. Simulations of k_snow using the snow physics model Crocus were performed. Contrary to observations, Crocus predicts high k_snow values at the base of the snowpack (0.12–0.27 W m⁻¹ K⁻¹) and low ones in its upper parts (0.02–0.12 W m⁻¹ K⁻¹). We diagnose that this is because Crocus does not describe the large upward water vapour fluxes caused by the temperature gradient in the snow and soil. These fluxes produce mass transfer between the soil and lower snow layers to the upper snow layers and the atmosphere. Finally, we discuss the importance of the structure and properties of the Arctic snowpack on subnivean life, as species such as lemmings live under the snow most of the year and must travel in the lower snow layer in search of food.

Accurately simulating the physical properties of Arctic snowpacks is essential for modeling the surface energy budget and the permafrost thermal regime. We show that the detailed snow physics models Crocus and SNOWPACK cannot simulate critical snow physical variables. Both models simulate basal layers with high density and high thermal conductivity, and top layers with low values for both variables, while field measurements yield opposite results. We explore the impact of an inverted snow stratigraphy on the permafrost thermal regime at a high Arctic site using a simplified heat transfer model and idealized snowpacks with three layers. One snowpack has a typical Arctic stratification with a low‐density insulating basal layer, while the other (called Alpine‐type snowpack) has a dense conducting basal layer. Snowpack stratification impacts simulated ground temperatures at 5 cm depth by less than 0.3 °C. Heat conduction through layered snowpacks is therefore determined by thermal insulance rather than by stratification. Ground dehydration caused by upward water vapor diffusion is 4 times greater under Arctic stratification, leading to a larger latent heat loss, but also to a lower soil thermal conductivity caused by ice loss, so that the overall effect of dehydration on ground temperature is uncertain. Snowpack stratification is found to affect snow surface temperature by up to 4 °C. Lastly, different snow metamorphism rates lead to a lower Alpine snowpack albedo, contributing to a warmer ground. Quantifying all these effects is needed for adequately simulating permafrost temperature. This requires the development of a snow and soil model that describes water vapor fluxes.

In the SURFEX/ISBA-Crocus multi-layer snowpack model, the snow microstructure has up to now been characterised by the grain size and by semi-empirical shape variables which cannot be measured easily in the field or linked to other relevant snow properties. In this work we introduce a new formulation of snow metamorphism directly based on equations describing the rate of change of the optical diameter (dopt). This variable is considered here to be equal to the equivalent sphere optical diameter, which is inversely proportional to the specific surface area (SSA). dopt thus represents quantitatively some of the geometric characteristics of a porous medium. Different prognostic rate equations of dopt, including a re-formulation of the original Crocus scheme and the parameterisations from Taillandier et al. (2007) and Flanner and Zender (2006), were evaluated by comparing their predictions to field measurements carried out at Summit Camp (Greenland) in May and June 2011 and at Col de Porte (French Alps) during the 2009/10 and 2011/12 winter seasons. We focused especially on results in terms of SSA. In addition, we tested the impact of the different formulations on the simulated density profile, the total snow height, the snow water equivalent (SWE) and the surface albedo. Results indicate that all formulations perform well, with median values of the RMSD between measured and simulated SSA lower than 10 m2 kg−1. Incorporating the optical diameter as a fully fledged prognostic variable is an important step forward in the quantitative description of the snow microstructure within snowpack models, because it opens the way to data assimilation of various electromagnetic observations.

With climate warming, shrubs have been observed to grow on Arctic tundra. Their presence is known to increase snow height and is expected to increase the thermal insulating effect of the snowpack. An important consequence would be the warming of the ground, which will accelerate permafrost thaw, providing an important positive feedback to warming. At Bylot Island (73° N, 80° W) in the Canadian high Arctic where bushes of willows (Salix richardsonii Hook) are growing, we have observed the snow stratigraphy and measured the vertical profiles of snow density, thermal conductivity and specific surface area (SSA) in over 20 sites of high Arctic tundra and in willow bushes 20 to 40 cm high. We find that shrubs increase snow height, but only up to their own height. In shrubs, snow density, thermal conductivity and SSA are all significantly lower than on herb tundra. In shrubs, depth hoar which has a low thermal conductivity was observed to grow up to shrub height, while on herb tundra, depth hoar only developed to 5 to 10 cm high. The thermal resistance of the snowpack was in general higher in shrubs than on herb tundra. More signs of melting were observed in shrubs, presumably because stems absorb radiation and provide hotspots that initiate melting. When melting was extensive, thermal conductivity was increased and thermal resistance was reduced, counteracting the observed effect of shrubs in the absence of melting. Simulations of the effect of shrubs on snow properties and on the ground thermal regime were made with the Crocus snow physics model and the ISBA (Interactions between Soil–Biosphere–Atmosphere) land surface scheme, driven by in situ and reanalysis meteorological data. These simulations did not take into account the summer impact of shrubs. They predict that the ground at 5 cm depth at Bylot Island during the 2014–2015 winter would be up to 13 °C warmer in the presence of shrubs. Such warming may however be mitigated by summer effects.

Basal depth hoar that forms in Arctic snowpacks often has a low thermal conductivity, strongly contributing to the snowpack thermal insulance and impacting the permafrost thermal regime. At Ward Hunt Island (Canadian high Arctic, 83°05′N, 74°07′W) almost no depth hoar was observed in spring 2016 despite favorable thermal conditions. We hypothesize that depth hoar formation was impeded by the combination of two factors (1) strong winds in fall that formed hard dense wind slabs where water vapor transport was slow and (2) low soil moisture that led to rapid ground cooling with no zero-curtain period, which reduced soil temperature and the temperature gradient in the snowpack. Comparisons with detailed data from the subsequent winter at Ward Hunt and from Bylot Island (73°09′N, 80°00′W) and with data from Barrow and Alert indicate that both high wind speeds after snow onset and low soil moisture are necessary to prevent Arctic depth hoar formation. The role of convection to form depth hoar is discussed. A simple preliminary strategy to parameterize depth hoar thermal conductivity in snow schemes is proposed based on wind speed and soil moisture. Finally, warming-induced vegetation growth and soil moisture increase should reduce depth hoar thermal conductivity, potentially affecting permafrost temperature.

The broadband albedo of surface snow is determined both by the near-surface profile of the physical and chemical properties of the snowpack and by the spectral and angular characteristics of the incident solar radiation. Simultaneous measurements of the physical and chemical properties of snow were carried out at Summit Camp, Greenland (72°36´ N, 38°25´ W, 3210 m a.s.l.) in May and June 2011, along with spectral albedo measurements. One of the main objectives of the field campaign was to test our ability to predict snow spectral albedo by comparing the measured albedo to the albedo calculated with a radiative transfer model, using measured snow physical and chemical properties. To achieve this goal, we made daily measurements of the snow spectral albedo in the range 350–2200 nm and recorded snow stratigraphic information down to roughly 80 cm. The snow specific surface area (SSA) was measured using the DUFISSS instrument (DUal Frequency Integrating Sphere for Snow SSA measurement, Gallet et al., 2009). Samples were also collected for chemical analyses including black carbon (BC) and dust, to evaluate the impact of light absorbing particulate matter in snow. This is one of the most comprehensive albedo-related data sets combining chemical analysis, snow physical properties and spectral albedo measurements obtained in a polar environment. The surface albedo was calculated from density, SSA, BC and dust profiles using the DISORT model (DIScrete Ordinate Radiative Transfer, Stamnes et al., 1988) and compared to the measured values. Results indicate that the energy absorbed by the snowpack through the whole spectrum considered can be inferred within 1.10%. This accuracy is only slightly better than that which can be obtained considering pure snow, meaning that the impact of impurities on the snow albedo is small at Summit. In the near infrared, minor deviations in albedo up to 0.014 can be due to the accuracy of radiation and SSA measurements and to the surface roughness, whereas deviations up to 0.05 can be explained by the spatial heterogeneity of the snowpack at small scales, the assumption of spherical snow grains made for DISORT simulations and the vertical resolution of measurements of surface layer physical properties. At 1430 and around 1800 nm the discrepancies are larger and independent of the snow properties; we propose that they are due to errors in the ice refractive index at these wavelengths. This work contributes to the development of physically based albedo schemes in detailed snowpack models, and to the improvement of retrieval algorithms for estimating snow properties from remote sensing data.