On the treatment of soil water stress in LSM simulations of vegetation function

Reliable estimates of ecosystem responses to climate variability and weather extremes depend on a realistic description of land surface-atmosphere feedbacks in environmental (weather and climate) prediction models. These comprise coupled photosynthesis–stomatal conductance (A–gs) sub-models, embedded in land surface models (LSMs), to estimate terrestrial fluxes of energy, water and carbon. LSMs incorporate parameterisations of environmental limitations, enabling realistic feedbacks; soil moisture limitation of plant function is a primary control on feedbacks, particularly in sub-tropical to temperate ecosystems. The representation of this process differs greatly among LSMs. Following Egea et al. (2011), we have implemented higher levels of biophysical complexity in the A-gs model embedded in the JULES LSM, by allowing root zone soil moisture to limit plant function via three individual routes (biochemical, stomatal conductance and mesophyll conductance) and combinations thereof. We performed land surface climate simulations, in the form of systematic sensitivity studies, over a large European domain for the period 1990 to 2009. We show, also by comparison with aggregated FLUXNET observations of Evapotranspiration and GPP, that current approaches to the treatment of soil water stress in LSMs fail to credibly simulate vegetation response across the typical range of European weather and climate variability; this behaviour is particularly evident in recent climate extremes, like the summer of 2003. Vegetation models retaining stomatal and mesophyll mechanisms in the imposition of soil water stress on plants are better able to discriminate the responses of water-limited versus radiation-limited regions. The impact on soil water resources is very significant, as well as regional feedbacks: a higher simulated water use efficiency in Southern Europe (with ~20% less soil water depletion for equivalent photosynthetic rates) supports vegetation recovery within a year of each extreme summer, in agreement with observations, unlike the standard (biochemical route) LSM.