Improving fractional snow cover retrieval from passive microwave data using a radiative transfer model and machine learning method

Abstract. The energy budget of Arctic sea ice is strongly affected by the snow cover. Intensive sampling of snow properties was conducted near Qikiqtarjuak in Baffin Bay on typical landfast sea ice during two melt seasons in 2015 and 2016. The sampling included stratigraphy, vertical profiles of snow specific surface area (SSA), density and surface spectral albedo. Both seasons feature four main phases: I) dry snow cover, II) surface melting, III) ripe snowpack and IV) melt pond formation. Each of them was characterized by distinctive physical and optical properties. Highest SSA of 49.3 m2 kg−1 was measured during phase I on surface windslab together with a high broadband albedo of 0.87. The next phase was marked by alternative episodes of surface melting which dramatically decreased the SSA below 3 m2 kg−1 and episodes of snowfall reestablishing the pre-melt conditions. Albedo was highly time variable especially in the near-infrared with minimum values around 0.45 at 1000 nm. At some point, the melt progressed leading to a fully ripe snowpack composed of clustered rounded grains in phase III. Albedo began to decrease in the visible as snow thickness decreased but remained steady at longer wavelengths. Moreover, its spatial variability clearly appeared for the first time following snow depth heterogeneity. The impacts on albedo of both snow SSA and thickness were quantitatively investigated using a radiative transfer model. Comparisons between albedo measurements and simulations show that our data on snow physical properties are relevant for radiative transfer modeling. They also point out to the importance of the properties of the very surface snow layer for albedo computation, especially during phase II when several distinctive layers of snow superimposed following snowfalls, melt or diurnal cycles.

Download Full-text

Towards improving radiometry-derived snow depths with SNOWPACK and SMRT

10.5194/egusphere-egu2020-19061 ◽

2020 ◽

Author(s):

Robbie Mallett ◽

Julienne Stroeve ◽

Michel Tsamados ◽

Glen Liston

Keyword(s):

Radiative Transfer ◽

Sea Ice ◽

Model Development ◽

Light Flux ◽

Passive Microwave ◽

Microwave Emission ◽

Radiative Transfer Model ◽

Transfer Model ◽

Gradient Ratio ◽

Strong Control

The depth of overlying snow on sea ice exerts a strong control on atmosphere-ocean heat and light flux and introduces major uncertainties in the remote sensing of sea ice thickness. Satellite-mounted microwave radiometers have enabled retrieval of snow depths over first year ice, but such retrievals are subject to a wide margin of error due to spatial variation in snow stratigraphy and roughness.Here we model the microwave signature of snow on sea ice using a recently released sea ice variant of the snowpack evolution model, SNOWPACK (Wever et al., 2020). By advecting parcels of sea ice using ice motion vectors and exposing them to the relevant atmospheric forcing using ERA5 reanalysis, we model the accumulation of snow and the development of snowpack stratigraphy.We then pass these modelled snowpacks to the Snow Microwave Radiative Transfer model (Picard et al., 2018) to estimate their microwave emission characteristics. By using relationships from the literature relating the ratios of the 37GHz and 19GHz channels, we calculate whether the traditional &#8220;gradient ratio&#8221; method (Markus and Cavalieri, 1998) over- or underestimates the depth of snow at a particular point based on our modelling. We then adjust the observed gradient ratio based on the model results in an attempt to better characterise snow depths.&#160;ReferencesWever, Nander, et al. "Version 1 of a sea ice module for the physics-based, detailed, multi-layer SNOWPACK model." Geoscientific Model Development 13.1 (2020): 99-119.Picard, Ghislain, Melody Sandells, and Henning L&#246;we. "SMRT: An active&#8211;passive microwave radiative transfer model for snow with multiple microstructure and scattering formulations (v1. 0)." Geoscientific Model Development 11.7 (2018): 2763-2788.Markus, Thorsten, and Donald J. Cavalieri. "Snow depth distribution over sea ice in the Southern Ocean from satellite passive microwave data." Antarctic sea ice: physical processes, interactions and variability 74 (1998): 19-39.

Download Full-text