Incorporating dynamic ozone stress and parameter optimization into the BEPS model for improved GPP simulation

Accurate simulation of gross primary productivity (GPP) under environmental stress remains a key challenge in process-based ecosystem models. In many existing models, ozone effects are represented using simplified or static parameter adjustments, which limits their ability to capture both the physiological inhibition and cumulative structural responses of vegetation. In this study, we enhanced the Boreal Ecosystem Productivity Simulator (BEPS) by incorporating a dynamic ozone stress mechanism based on stomatal ozone flux and by optimizing key ecophysiological parameters using the Generalized Likelihood Uncertainty Estimation (GLUE) approach, resulting in an improved model configuration (BEPS_OP). The performance of the original BEPS, an ozone-integrated version (BEPS_O3), and BEPS_OP was evaluated using 8-day gross primary productivity observations from multiple FLUXNET2015 eddy covariance sites. Model simulations were further compared with independent remote sensing GPP products, including MODIS GPP and the TL-LUE dataset. Results indicate that the explicit representation of ozone stress combined with parameter optimization substantially reduced simulation errors and improved the temporal consistency of GPP estimates across diverse ecosystems, with particularly strong performance in forest and shrubland systems, as evidenced by substantial reductions in RMSE and MAE. Analysis of ozone impacts reveals systematic GPP reductions under elevated ozone exposure, with distinct sensitivity patterns among vegetation types. Overall, this study demonstrates that enhancing the mechanistic representation of ozone-related processes and accounting for parameter uncertainty can improve the robustness and ecological realism of GPP simulations, providing a practical pathway for strengthening carbon cycle modeling under air pollution stress.