Skip to main content
U.S. flag

An official website of the United States government

Metrics, Measurements, and Methods for Advancing Ecosystem Model Evaluation

Presentation Date
Friday, December 17, 2021 at 12:32pm
Location
Convention Center - Room 231-232
Authors

Author

Abstract

The mechanistic representation of biogeochemical processes in ecosystem models is rapidly advancing, requiring similar advancements in model evaluation approaches. Model benchmarking techniques need to evaluate functional relationships among ecosystem processes, particularly when ecosystem models are used to make projections under changing climate conditions with no historical analog. Here we leverage site-level ecosystem measurements to assess model representations of ecosystem functioning using dynamic response metrics, underutilized observational measurements, and new analytical methods. First, we examined modeled ecosystem response to precipitation pulse events to assess hydraulic functioning on short time scales. This dynamic response metric was a particularly salient test for evaluating plant hydraulic functioning and modeled ecohydrological cycling. Secondly, we used emerging new measurements of atmospheric carbon isotope concentrations to examine simulated water-use efficiency on diurnal time scales. These measurements shed new light on stomatal functioning on short time scales and can be used to evaluate model representation of stomatal functioning under environmental stress. Lastly, we used information theory to examine the functional relationships among soil water potential, vapor pressure deficit, and transpiration in both observations and model simulations. This technique quantified both model predictive and functional performance and attributed errors to specific processes making it a powerful tool for model evaluation. Collectively these methods illustrate next directions for model benchmarking, highlighting new measurements and techniques for evaluating model representation of ecosystem functioning.

Category
Biogeosciences
Funding Program Area(s)