Modeling of interaction of electromagnetic (EM) waves in the optical spectrum with sea ice, snow, and melt ponds

Optical properties of the sea ice surface are highly variable both in space and time. The surface of Arctic sea ice is determined by snow and ice, and during summer also by melt ponds. The optical properties of sea ice are important to know as they are key factors in the energy balance and climate feedbacks, in particular, the ice-albedo feedback (Curry et al., 1996). Besides that, the surface optical properties such as albedo and bidirectional reflectance distribution function (BRDF) are connected to a variety of macro- and microphysical parameters of the surface (e.g., snow grain size, soot content, ice thickness). Comprehensive field measurements with sufficient spatial and temporal coverage are challenging due to the harsh conditions and remoteness of the area. To improve our understanding of the system "ocean-sea ice-atmosphere", remote sensing retrievals and modeling have to be utilized in addition to field measurements.

There are two satellite retrievals of the optical properties of the Arctic sea ice surface currently operational in the group: the Snow Grain Size and Pollution (SGSP) retrieval based on MODIS data (Zege et al., 2011; Wiebe et al, 2013) and the Melt Pond Detector (MPD) retrieval using MERIS/OLCI data (Zege et al., 2015; Istomina et al., 2015a,b). These retrievals are used to estimate the grain size, soot content and spectral albedo of snow (SGSP), and spectral albedo and melt pond fraction of the Arctic sea ice (MPD). Both retrievals are based on a forward model of light scattering within snow/sea ice. The version of the forward model for snow and ice used for the retrieval is described in Malinka et al., 2016 (see also SIDARUS Deliverable D4.4 [PDF]). It is based on an analytical approximation that has been developed on the basis of the asymptotic solution of the radiative transfer theory (Zege et al., 1991). The model describes the medium as a distribution of chord lengths rather than as a set of separate particles with known properties. The mean chord length equals the mean photon path length inside one of the components of the medium. For the wavelengths in the visible spectral range and the typical grain sizes of snow and sea ice (greater than 100µm) the laws of geometrical optics are applicable. The BRDF of snow and ice is determined by the optical depth of the surface, mean effective grain size and potentially particulate matter or contaminants in the snow/ice. The BRDF of a melt pond is calculated as a combination of the melt water component given by the optical depth of the water layer, and the pond bottom component, which as determined by the optical depth of the underlying sea ice. Frozen over melt ponds can also be accounted for.

The forward model has been extensively validated against field measurements of spectral albedo of the Arctic sea ice and melt ponds, and showed good correspondence and robustness (Malinka et al., 2016)

The optical model of sea ice and melt ponds has been utilized to derive melt pond fraction and sea ice albedo for the historic MERIS dataset 2002-2011 published on this website.

Current work focuses on including open water, like it occurs in leads and polynyas within the sea ice, as an additional surface class and on a better description of over frozen ponds within the forward model.


Figure 1: Comparison of the spectral albedo measured in the field (blue curve) to forward modeling (green curve) for (left) dry white sea ice at 58° illumination under overcast skies with solar disk visible and (right) for a dark frozen over melt pond under same illumination and sky conditions. The field measurements have been performed during the RV Polarstern cruise PS 89/3 in the Central Arctic in 2012 (see Field Datasets).