An observational constraint on stomatal function in forests: evaluating coupled carbon and water vapor exchange with carbon isotopes in the Community Land Model (CLM4.5)

Abstract. Land surface models are useful tools to quantify contemporary and future climate impact on terrestrial carbon cycle processes, provided they can be appropriately constrained and tested with observations. Stable carbon isotopes of CO₂ offer the potential to improve model representation of the coupled carbon and water cycles because they are strongly influenced by stomatal function. Recently, a representation of stable carbon isotope discrimination was incorporated into the Community Land Model component of the Community Earth System Model. Here, we tested the model's capability to simulate whole-forest isotope discrimination in a subalpine conifer forest at Niwot Ridge, Colorado, USA. We distinguished between isotopic behavior in response to a decrease of δ¹³C within atmospheric CO₂ (Suess effect) vs. photosynthetic discrimination (Δ_canopy), by creating a site-customized atmospheric CO₂ and δ¹³C of CO₂ time series. We implemented a seasonally varying V_cmax model calibration that best matched site observations of net CO₂ carbon exchange, latent heat exchange, and biomass. The model accurately simulated observed δ¹³C of needle and stem tissue, but underestimated the δ¹³C of bulk soil carbon by 1–2 ‰. The model overestimated the multiyear (2006–2012) average Δ_canopy relative to prior data-based estimates by 2–4 ‰. The amplitude of the average seasonal cycle of Δ_canopy (i.e., higher in spring/fall as compared to summer) was correctly modeled but only when using a revised, fully coupled A_n − g_s (net assimilation rate, stomatal conductance) version of the model in contrast to the partially coupled A_n − g_s version used in the default model. The model attributed most of the seasonal variation in discrimination to A_n, whereas interannual variation in simulated Δ_canopy during the summer months was driven by stomatal response to vapor pressure deficit (VPD). The model simulated a 10 % increase in both photosynthetic discrimination and water-use efficiency (WUE) since 1850 which is counter to established relationships between discrimination and WUE. The isotope observations used here to constrain CLM suggest (1) the model overestimated stomatal conductance and (2) the default CLM approach to representing nitrogen limitation (partially coupled model) was not capable of reproducing observed trends in discrimination. These findings demonstrate that isotope observations can provide important information related to stomatal function driven by environmental stress from VPD and nitrogen limitation. Future versions of CLM that incorporate carbon isotope discrimination are likely to benefit from explicit inclusion of mesophyll conductance.