Abstract
Abstract. The response function identification method introduced in the first part of this study is applied here to investigate the land carbon cycle in the Max Planck Institute for Meteorology Earth System Model. We identify from standard C4MIP 1 % experiments the linear response functions that generalize the land carbon sensitivities β and γ. The identification of these generalized sensitivities is shown to be robust by demonstrating their predictive power when applied to experiments not used for their identification. The linear regime for which the generalized framework is valid is estimated, and approaches to improve the quality of the results are proposed. For the generalized γ sensitivity, the response is found to be linear for temperature perturbations until at least 6 K. When this sensitivity is identified from a 2×CO2 experiment instead of the 1 % experiment, its predictive power improves, indicating an enhancement in the quality of the identification. For the generalized β sensitivity, the linear regime is found to extend up to CO2 perturbations of 100 ppm. We find that nonlinearities in the β response arise mainly from the nonlinear relationship between net primary production and CO2. By taking as forcing the resulting net primary production instead of CO2, the response is approximately linear until CO2 perturbations of about 850 ppm. Taking net primary production as forcing also substantially improves the spectral resolution of the generalized β sensitivity. For the best recovery of this sensitivity, we find a spectrum of internal timescales with two peaks, at 4 and 100 years. Robustness of this result is demonstrated by two independent tests. We find that the two-peak spectrum can be explained by the different characteristic timescales of functionally different elements of the land carbon cycle. The peak at 4 years results from the collective response of carbon pools whose dynamics is governed by fast processes, namely pools representing living vegetation tissues (leaves, fine roots, sugars, and starches) and associated litter. The peak at 100 years results from the collective response of pools whose dynamics is determined by slow processes, namely the pools that represent the wood in stem and coarse roots, the associated litter, and the soil carbon (humus). Analysis of the response functions that characterize these two groups of pools shows that the pools with fast dynamics dominate the land carbon response only for times below 2 years. For times above 25 years the response is completely determined by the pools with slow dynamics. From 100 years onwards only the humus pool contributes to the land carbon response.
Highlights
In Part 1 of this study (Torres Mendonca et al, 2021a) we developed a method to identify linear response functions from arbitrary perturbation experiments
Torres Mendonça et al.: Identification of linear response functions – Part 2 vestigate the dynamics of the land carbon cycle in the Max Planck Institute for Meteorology Earth System Model (MPIESM; see Appendix A)
In preparation for future studies applying these generalized sensitivities to study the dynamics of the carbon climate system in C4MIP models, we investigate various ideas to improve the quality of the recovery of the response functions by using additional types of data routinely available in C4MIP simulations or using log-transformed data to account at least partially for process-immanent nonlinearities that hinder the usage of experiment data from larger levels of forcing
Summary
In Part 1 of this study (Torres Mendonca et al, 2021a) we developed a method to identify linear response functions from arbitrary perturbation experiments. In preparation for future studies applying these generalized sensitivities to study the dynamics of the carbon climate system in C4MIP models, we investigate various ideas to improve the quality of the recovery of the response functions by using additional types of data routinely available in C4MIP simulations or using log-transformed data to account at least partially for process-immanent nonlinearities that hinder the usage of experiment data from larger levels of forcing. In the present work we show that, in contrast to results obtained with classical fitting procedures, spectra recovered by the RFI method may be reliable and even interpretable For this purpose, we investigate a relatively detailed spectrum of timescales that arises from a high-quality recovery of the generalized β sensitivity.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.