“…Even though many applications still rely on empirical approaches to relate snowpack properties (e.g., snow water equivalent, SWE) and measured signals, it is generally accepted that a physical understanding of the interaction between snow and electromagnetic waves is necessary to improve the accuracy and overcome inherent difficulties of the retrieval as an underdetermined problem. The retrieval of snow properties is therefore often preceded by forward modeling and data as-similation (Durand and Margulis, 2007;Picard et al, 2009;Takala et al, 2011;Toure et al, 2011;Huang et al, 2012) to predict the satellite signal from prescribed snowpack properties that can be either obtained from measurements (e.g., Rosenfeld and Grody, 2000;Brucker et al, 2011a;Rees et al, 2010;Derksen et al, 2012Derksen et al, , 2014Kontu et al, 2014) or snow models (e.g., Flach et al, 2005;Brucker et al, 2011b;Andreadis and Lettenmaier, 2012;Kang and Barros, 2012;Wójcik et al, 2008;Kontu et al, 2017). The actual modeling challenge lies in the snowpack and the underlying surface (soil, ice, or water) where the coupling of various ingredients needs to be understood with sufficient accuracy to build efficient forward models.…”