Temporally variable parameters simulate asymmetrical interannual variation of aboveground and belowground carbon pools in an alpine meadow

Abstract Exploring interannual variation (IAV) of terrestrial carbon cycle is crucial for understanding the climate-carbon cycle feedback. However, model simulations of IAV in carbon pools are highly sensitive to parameterization. Although researches indicate that parameters vary as environment changes, most terrestrial carbon cycle models apply time-invariant parameters, leading to uncertainty in reproducing IAV. We compared the IAV simulation accuracy of carbon pools between constant and temporally variable parameters, generated by a piecewise model-data fusion framework based on long-term multi-observations in an alpine meadow over the Tibetan Plateau. Piecewise assimilation showed that key parameters involved in photosynthesis and allocation processes exhibited significant temporal variation under warming and nitrogen deposition. Specifically, day length threshold for leaf senescence was postponed, while allocation to autotrophic respiration, allocation to root, leaf nitrogen content, and nitrogen use efficiency increased. Compared with constant parameters, temporally variable parameters captured the observed asymmetrical response of aboveground and belowground carbon pools to climate fluctuations, improved modeled magnitude of IAV in carbon pools by 34% on average, and advanced carbon sink estimation accuracy by 43.15%. Therefore, fitting parameter dynamics to climatic conditions can provide a benchmark for future simulation in the Tibetan Plateau. Our research suggests that IAV in different carbon pools is crucial in determining carbon sinks. Temporal variations of parameters present how ecosystems adapt to a changing environment and should be incorporated to terrestrial carbon cycle models to reduce uncertainty in model prediction.