“…These satellite-based SWE retrieval models are often affected by errors arising from meteorological fields (e.g., data aggregation, disaggregation, extrapolation and interpolation (Blöschl & Sivapalan, 1995)) used to force land surface models. They are also affected by significant uncertainties associated with snow stratigraphy (Derksen, Walker, & Goodison, 2005), snow grain size (Armstrong, Chang, Rango, & Josberger, 1993), depth hoar layer (Brucker, Royer, Picard, Langlois, & Fily, 2011;Hall, 1987;Hall, Chang, & Foster, 1986;Foster et al, 2005), ice crusts (Rees, Lemmetyinen, Derksen, Pulliainen, & English, 2010), lake fraction effects , and snow morphology (Kelly et al, 2003), especially in densely-vegetated regions (Tedesco & Narvekar, 2010;Derksen et al, 2005) with relatively deep snow (Clifford, 2010). Remote Sensing of Environment 170 (2015) [153][154][155][156][157][158][159][160][161][162][163][164][165] In an effort to overcome many of the limitations highlighted above, the third alternative involves merging measurements of remote sensing observations with estimates from physically-based models (Reichle, 2008;Forman, Reichle, & Rodell, 2012;Reichle, De Lannoy, Forman, Draper, & Liu, 2014) using data assimilation (DA).…”