“…This has received substantial attention in the land modeling community over the past several decades [Shuttleworth, 1988;Entekhabi and Eagleson, 1989;Pitman et al, 1990;Dolman and Gregory, 1992;Koster and Suarez, 1992], and has been addressed in five main ways: (1) explicit representation of subgrid variability-this can be accomplished by configuring the land model with a finer mesh than the rest of the ESM [Hahmann and Dickinson, 2001], by using multiple tiles to represent subgrid heterogeneity [Koster and Suarez, 1992;Bonan et al, 2002], or by explicitly representing the spatial variability for a subset of processes; e.g., separate stomatal conductance calculations for sunlit and shaded leaves [Wang and Leuning, 1998] or separate energy balance calculations for snow-covered and snow-free surfaces [Takata et al, 2003;; (2) statistical-dynamical models, which parameterize how subgrid variability in model state variables affects grid-average fluxes-for example, as discussed in section A1.1, the Probability Distributed Model [Moore and Clarke, 1981] and TOPMODEL [Beven and Kirkby, 1979] represent the impacts of A key missing link in the current generation of land models is representing how the spatial organization of soil moisture and groundwater [Western et al, 1999;Grant et al, 2004] affects land-atmosphere fluxes [Maxwell and Kollet, 2008b]. In particular, most of the land models reviewed in Table 2 have a simplistic representation of the topographic controls on fine-scale soil moisture heterogeneity and the associated heterogeneity in evapotranspiration.…”