Plant water-stress parameterization determines the strength of land-atmosphere coupling

Abstract Land-surface models used in studies of the atmosphere and vegetation during droughts usually include an underlying parameterization that describes the response of plants to water stress. Here, we show that different formulations of this parameterization can lead to significant differences in the coupling strength (i.e. the magnitude of the carbon and water exchange) between the land surface and the atmospheric boundary layer (ABL). We use a numerical model that couples the daytime surface fluxes typical for low vegetation to the dynamics of a convective ABL, to systematically investigate a range of plant water-stress responses. We find that under dry soil conditions, changing from a sensitive to an insensitive vegetation response to water stress has the same impact on the land–atmosphere (L–A) coupling as a strong increase in soil moisture content. The insensitive vegetation allows stomata to remain open for transpiration (+150 W m −2 compared to the sensitive one), which cools the atmosphere (−3.5 K) and limits the ABL growth (−500 m). During the progressive development of a dry spell, the insensitive response will first dampen atmospheric heating because the vegetation continues to transpire a maximum of 4.6 mm day −1 while soil moisture is available. In contrast, the more sensitive vegetation response reduces its transpiration by more than 1 mm day −1 to prevent soil moisture depletion. But when soil moisture comes close to wilting point, the insensitive vegetation will suddenly close its stomata causing a switch to a L–A coupling regime dominated by sensible heat exchange. We find that in both cases, progressive soil moisture depletion contributes to further atmospheric warming up to 6 K, reduced photosynthesis up to 89%, and CO 2 enrichment up to 30 ppm, but the full impact is strongly delayed for the insensitive vegetation. Then, when we analyze the impact of a deviation of the modeled large-scale boundary conditions (e.g. subsidence, cloud cover, free-troposphere lapse rates, etc.) from their true state during a drought, we find that the two coupled systems (with a sensitive or insensitive vegetation) respond much differently to the generated atmospheric warming, this due to the difference in the basic surface coupling regime (coupled vs. uncoupled). This is of importance for the simulation of heat waves and meteorological droughts, as well as carbon-climate projections, as we show the predictive skill of coupled models is tied to the underlying vegetation response to water stress.