Land surface conductance linked to precipitation: Co-evolution of vegetation and climate in Earth system models.

Vegetation and precipitation are known to fundamentally influence each other. However, this interdependence is not fully represented in climate models because the characteristics of land surface (canopy) conductance to water vapor and CO2 are determined independently of precipitation. Working within a coupled atmosphere and land modelling framework (CAM6/CLM5; coupled Community Atmosphere Model v6/Community Land Model v5), we have developed a new theoretical approach to characterizing land surface conductance by explicitly linking its dynamic properties to local precipitation, a robust proxy for moisture available to vegetation. This will enable regional surface conductance characteristics to shift fluidly with climate change in simulations, consistent with general principles of co-evolution of vegetation and climate. Testing within the CAM6/CLM5 framework shows that climate simulations incorporating the new theory outperform current default configurations across several error metrics for core output variables when measured against observational data. In climate simulations for the end of this century the new, adaptive stomatal conductance scheme provides a revised prognosis for average and extreme temperatures over several large regions, with increased primary productivity through central and east Asia, and higher rainfall through North Africa and the Middle East. The new projections also reveal more frequent heatwaves than originally estimated for the south-eastern US and sub-Saharan Africa but less frequent heatwaves across east Europe and northeast Asia. These developments have implications for evaluating food security and risks from extreme temperatures in areas that are vulnerable to climate change.

Comparing optimal and empirical stomatal conductance models for application in Earth system models.

Earth system models (ESMs) rely on the calculation of canopy conductance in land surface models (LSMs) to quantify the partitioning of land surface energy, water, and CO2 fluxes. This is achieved by scaling stomatal conductance, gw , determined from physiological models developed for leaves. Traditionally, models for gw have been semi-empirical, combining physiological functions with empirically determined calibration constants. More recently, optimization theory has been applied to model gw in LSMs under the premise that it has a stronger grounding in physiological theory and might ultimately lead to improved predictive accuracy. However, this premise has not been thoroughly tested. Using original field data from contrasting forest systems, we compare a widely used empirical type and a more recently developed optimization-type gw model, termed BB and MED, respectively. Overall, we find no difference between the two models when used to simulate gw from photosynthesis data, or leaf gas exchange from a coupled photosynthesis-conductance model, or gross primary productivity and evapotranspiration for a FLUXNET tower site with the CLM5 community LSM. Field measurements reveal that the key fitted parameters for BB and MED, g1B and g1M, exhibit strong species specificity in magnitude and sensitivity to CO2 , and CLM5 simulations reveal that failure to include this sensitivity can result in significant overestimates of evapotranspiration for high-CO2 scenarios. Further, we show that g1B and g1M can be determined from mean ci /ca (ratio of leaf intercellular to ambient CO2 concentration). Applying this relationship with ci /ca values derived from a leaf δ13 C database, we obtain a global distribution of g1B and g1M , and these values correlate significantly with mean annual precipitation. This provides a new methodology for global parameterization of the BB and MED models in LSMs, tied directly to leaf physiology but unconstrained by spatial boundaries separating designated biomes or plant functional types.