Photosynthesis and Phenology Parameter Optimization Alleviates Terrestrial Biosphere Model Underestimate of Net CO2 Flux Interannual Variability at Semiarid Sites

Abstract:

Recent modeling studies have shown that semiarid ecosystems play a dominant role in the interannual variability (IAV) of the global carbon (C) sink. However, the global terrestrial biosphere models (TBMs) used in these studies have not yet been extensively tested or optimized against semi-arid field data. Recent research comparing a suite of TBMs to semi-arid site CO₂ flux data in the southwestern (SW) US showed that all models underestimate both the mean C source/sink behavior and net CO₂ flux IAV. Given the lack of model parameter calibration studies that have included semiarid sites in their optimizations, it remains to be seen whether these model-data discrepancies are due to uncertain model processes or inaccurate parameters. To bridge this gap, we tested whether parameter optimization could alleviate the SW US semiarid site model-data discrepancies in the mean C source/sink patterns and net CO₂ flux IAV. Using a Bayesian data assimilation framework, we optimized photosynthesis, phenology, C allocation and respiration parameters of the ORCHIDEE TBM with CO₂ flux data from 12 SW US Ameriflux sites spanning forest, shrub and grassland semiarid ecosystems. Across all sites, we found that parameter optimization dramatically improves the model’s ability to capture both the mean C source/sink behavior as well as net CO₂ flux IAV. By performing multiple optimization tests with different combinations of parameters, we were able to identify the physiological processes that were most influential in improving the model behavior. Overall, photosynthesis parameters were crucial for reducing the bias in the mean annual C budget (therefore, the source vs sink behavior of a site) as well as for improving the magnitude and sign of the IAV. In addition to photosynthesis, phenology parameters were also important for capturing the IAV by correcting the timing and magnitude of peak gross CO₂ uptake. Optimizing C allocation and respiration parameters was needed to capture the correct annual CO₂ budget for the strongest C sink sites. Overall, this study demonstrated that in addition to further development of semi-arid C and vegetation related processes, modeling groups need to perform parameter optimizations in order to achieve reliable estimates of semi-arid ecosystem contributions to the global C cycle.