New Mechanisms, Feedbacks, and Approaches for Improving Predictions of Global Biogeochemical Cycles in Earth System Models II
Friday, December 16, 2016 17:00–17:15
Moscone West 2020
Since nutrient limitations of the terrestrial carbon cycle have been rigorously demonstrated by field experiments (e.g., fertilization experiments), various Earth System Model (ESM) groups have spent substantial effort to incorporate nutrient sub-models in their terrestrial carbon cycle models (e.g., CLM-CNP, JSBACH-CNP, CABLE-CNP).
Here we report new developments in the Accelerated Climate Modeling for Energy (ACME) Land Model (ALMv1) regarding carbon-nutrient interactions. The development is based on (1) recent theoretical advances in understanding belowground multiple-consumer, multiple-nutrient competition; (2) a dynamic allocation scheme based on resources availability to balance whole system functioning; and (3) global datasets of plant physiological traits (e.g., TRY database). In addition to describing these developments, we compare them with representations in other ESM land models.
We demonstrate that these new ALMv1 developments accurately reproduce present terrestrial carbon dynamics under the constraints of nitrogen and phosphorus availability. We benchmark the new model using the International Land Model Benchmarking package (ILAMB) for major plant and soil carbon pools (total vegetation biomass, soil carbon content) and fluxes (GPP, NEP) and leaf area index (LAI). The results show that biases have been significantly reduced (e.g., GPP bias in boreal needle leaf forests, LAI bias in tropical broad leaf evergreen forests). The ILAMB benchmarks provide a unique opportunity to consistently track ongoing model development and simultaneously demonstrate improvements for multiple variables.
We further benchmark the new model transient response to nutrient perturbation, using nitrogen and phosphorus fertilization experiments at over 100 forest sites. We found that the plant productivity response ratio ((NPPfert - NPPcontrol)/ NPPcontrol) could be captured only when the whole system functional balance is taken into account. Compared with other candidate models (e.g., fixed resource allocation), our results highlight the importance of self-regulation and adjustment of forest ecosystems in response to long-term resource supply imbalances (e.g., by CO2, N, P) that are likely to occur over the next several decades.