Assessing Ecosystem Drought Response in CLM 4.5 Using Site-Level Flux and Carbon-Isotope Measurements: Results From a Pacific Northwest Coniferous Forest

Abstract:

The frequency, extent, and severity of droughts are expected to increase in the western United States as climate changes occur. The combination of warmer temperature, larger vapor pressure deficit, reduced snowfall and snow pack, earlier snow melt, and extended growing seasons is expected to lead to an intensification of summer droughts, with a direct impact on ecosystem productivity and therefore on the carbon budget of the region. In this scenario, an accurate representation of ecosystem drought response in land models becomes fundamental, but the task is challenging, especially in regards to stomatal response to drought.

In this study we used the most recent release of the Community Land Model (CLM 4.5), which now includes photosynthetic carbon isotope discrimination and revised photosynthesis and hydrology schemes, among an extensive list of updates. We evaluated the model’s performance at a coniferous forest site in the Pacific northwest (Wind River AmeriFlux Site), characterized by a climate that has a strong winter precipitation component followed by a summer drought. We ran the model in offline mode (i.e., decoupled from an atmospheric model), forced by observed meteorological data, and used site observations (e.g., surface fluxes, biomass values, and carbon isotope data) to assess the model.

Previous field observations indicated a significant negative correlation between soil water content and the carbon isotope ratio of ecosystem respiration (δ¹³C_R), suggesting that δ¹³C_R was closely related to the photosynthetic discrimination against ¹³CO₂ as controlled by stomatal conductance. We used these observations and latent-heat flux measurements to assess the modeled stomatal conductance values and their responses to extended summer drought. We first present the model results, followed by a discussion of potential CLM model improvements in stomatal conductance responses and in the representation of soil water stress (parameter β_t) that would more precisely incorporate features that would allow the model to correctly simulate field observations.