Use of an environmental filtering parameterization of stomatal conductance in the Catchment-CN land surface model

Droughts are expected to increase in frequency and severity with climate change, decreasing the amount of water available to plants for transpiration. Correctly predicting how changes in water availability will affect transpiration-and thus the carbon, water, and energy cycles-requires accurately modeling stomatal conductance in Land Surface Models (LSMs). A key challenge in modeling stomatal conductance is the parameterization of stomatal marginal water use efficiency (g1), a stomatal trait that reflects the sensitivity of stomatal conductance to changing water availability. LSMs generally parameterize g1 using plant functional types (PFTs) and assume that g1 is identical for all locations with the same PFT. However, numerous studies have shown that there is more variability in stomatal traits, such as g1, within PFTs than across PFTs. At the same time, a lack of direct observations of stomatal conductance across the globe limits the use of more detailed parameterization schemes. Here, we fill this gap by developing an environmental filtering (EF) scheme for parameterizing pixel-specific values of g1 in the Catchment-CN LSM across North America. In this approach, g1 is assumed to vary spatially depending on climatic conditions and canopy properties. It is then optimally scaled within Catchment-CN using a particle swarm optimization to reduce compensating errors with other Catchment-CN model biases. We first use EF to fit a simple linear relationship between microwave radiometry-constrained g1 estimates and environmental features such as mean climate and canopy height. Using this approach, we find that mean annual precipitation alone is a significantly better predictor of g1 than are PFTs. We then use this g1-precipitation relationship to determine a specific g1 value for each pixel within Catchment-CN. In this presentation, we will compare predicted evapotranspiration and other hydrologic variables between the EF-parameterized Catchment-CN and non-EF parameterized Catchment-CN. In sum, our EF-based approach establishes a method for sub-PFT parameterization of stomatal traits in LSMs to predict vegetation response to a changing climate.