A global function of climatic aridity accounts for soil moisture stress on carbon assimilation
Abstract. The coupling between carbon uptake and water loss through stomata implies that gross primary production (GPP) can be limited by soil water availability through reduced leaf area and/or reduced stomatal conductance. Vegetation and land-surface models typically assume that GPP is highest under well-watered conditions and apply a stress function to reduce GPP with declining soil moisture below a critical threshold, which may be universal or prescribed by vegetation type. It is unclear how well current schemes represent the water conservation strategies of plants in different climates. Here eddy-covariance flux data are used to investigate empirically how soil moisture influences the light-use efficiency (LUE) of GPP. Well-watered GPP is estimated using the P model, a first-principles LUE model driven by atmospheric data and remotely sensed green vegetation cover. Breakpoint regression is used to relate the daily value of the ratio β(θ) (flux-derived GPP/modelled well-watered GPP) to soil moisture, which is estimated using a generic water-balance model. Maximum LUE, even during wetter periods, is shown to decline with increasing climatic aridity index (AI). The critical soil-moisture threshold also declines with AI. Moreover, for any AI, there is a value of soil moisture at which β(θ) is maximized, and this value declines with increasing AI. Thus, ecosystems adapted to seasonally dry conditions use water more conservatively (relative to well-watered ecosystems) when soil moisture is high, but maintain higher GPP when soil moisture is low. An empirical non-linear function of AI expressing these relationships is derived by non-linear regression, and used to generate a β(θ) function that provides a multiplier for well-watered GPP as simulated by the P model. Substantially improved GPP simulation is shown during both unstressed and water-stressed conditions, compared to the reference model version that ignores soil-moisture stress, and to an earlier formulation in which maximum LUE was not reduced. This scheme may provide a step towards better-founded representations of carbon-water cycle coupling in vegetation and land-surface models.
Viewed (geographical distribution)