Conceptual aspects of a statistical‐dynamical approach to represent landscape subgrid‐scale heterogeneities in atmospheric models

Land‐surface characteristics play a key role in the partition of energy received at the Earth's surface and, as a result, have a major impact on the atmosphere. Yet the representation of land‐surface processes in atmospheric models is not realistic. Actual state‐of‐the‐art parameterizations of land surfaces do not account for the landscape heterogeneity found on the resolvable scale of these models and are based on a large amount of empirical constants that in practice can be difficult to estimate. An alternative parameterization based on a statistical‐dynamical approach is suggested here. With this approach, the most important characteristics of the soil‐plant‐atmosphere system that affect the partition of energy (e.g., plant stomatal conductance, soil humidity, surface roughness) would be represented by a probability density function (pdf) rather than by a single “representative” value. Typically, such pdf's are characterized by two to four parameters. A primary simplified version of this parameterization was used to estimate the land‐surface energy fluxes that are produced at the grid scale by various distributions of stomatal conductance under a broad range of environmental conditions. To demonstrate the potential of the approach, results were compared with the same fluxes calculated with a big leaf model using the mean stomatal conductance that corresponds to the distributions. Large absolute and relative differences are obtained between the two schemes for many combinations of stomatal conductance pdf's and environmental conditions. These differences are due mainly to the nonlinearity of the processes involved in the redistribution of the energy absorbed at the ground surface. These numerical experiments demonstrate the importance of accounting for landscape heterogeneity in the simulation of land‐surface energy fluxes, and demonstrate the potential benefits of adopting a statistical‐dynamical approach for the representation of land‐surface processes in atmospheric models.