Functional Unit Testing: An Idea for Evaluating Ideas in Terrestrial Biogeochemistry

Abstract:

Experimental observations of terrestrial biogeochemistry are often made at the scale of leaves and the rhizosphere or at spatial scales finer than that of the entire ecosystem. Classic examples include measurements of leaf-level photosynthesis and of soil respiration using soil chambers. Field experiments are often designed to test hypotheses at these finer scales or to test larger scale hypotheses linked to those finer-scale processes. In turn, functional representations of processes within a model are themselves scale-dependent hypotheses. They may be hypotheses that are broadly supported by experimental results or they may be more tentative. In either case, there is a need to evaluate the finer-scale hypotheses within the context of the model’s integrated representation of larger-scale higher-level behavior. For example, model results can indicate how a hypothesized process at the scale of fine roots might be expressed at the scale of whole stand evapotranspiration. These results can then inform the design of experiments and observations best suited for capturing that multi-scale behavior and dependency. However, it is often extremely difficult to extract results of finer scale hypotheses (functions) from fully integrated large-scale ecosystem models. Accordingly, we have developed a framework for extracting individual process representations from the Community Land Model (CLM) into modular units for functional testing. These modules encapsulate fine-scale hypotheses about terrestrial biogeochemistry. Results can be generated quickly with these modules and compared directly with the experimental observations at the appropriate scale. Alternative hypotheses can be quickly implemented and evaluated. We illustrate this concept with results from CLM and the Partitioning in Trees and Soils loblolly pine field experiment. Our findings suggest deficits in the model’s hypothesized relationship between photosynthesis and temperature and suggest a focus in subsequent experiments on temperature response functions. We discuss the broader implications of using the functional unit testing in evaluating hypotheses and other ideas about terrestrial biogeochemical processes, including how our approach strengthens mutual feedbacks between model and field experiment.