Improvement of transpiration estimation based on a two-leaf conductance-photosynthesis model with seasonal parameters...

Conductance-photosynthesis (G_s-A) models, accompanying with light use efficiency (LUE) models for calculating carbon assimilation, are widely used for estimating canopy stomatal conductance (G_s) and transpiration (T_c) under the two-leaf (TL) scheme. However, the key parameters of photosynthetic rate sensitivity (g_su and g_sh) and maximum LUE (ϵ_msu and ϵ_msh) are typically set to temporally constant values for sunlit and shaded leaves, respectively. This may result in T_c estimation errors, as it contradicts field observations.

In this study, the measured flux data from three temperate deciduous broadleaved forests (DBF) FLUXNET sites were adopted, and the key parameters of LUE and Ball-Berry models for sunlit and shaded leaves were calibrated within the entire growing season and each season, respectively. Then, the estimations of gross primary production (GPP) and T_c were compared between the two schemes of parameterization: (1) entire growing season-based fixed parameters (EGS) and (2) season-specific dynamic parameters (SEA).

Our results show a cyclical variability of ϵ_msu across the sites, with the highest value during the summer and the lowest during the spring. A similar pattern was found for g_su and g_sh, which showed a decrease in summer and a slight increase in both spring and autumn. Furthermore, the SEA model (i.e., the dynamic parameterization) better simulated GPP, with a reduction in root mean square error (RMSE) of about 8.0 ± 1.1% and an improvement in correlation coefficient (r) of 3.7 ± 1.5%, relative to the EGS model. Meanwhile, the SEA scheme reduced T_c simulation errors in terms of RMSE by 3.7 ± 4.4%.

These findings provide a greater understanding of the seasonality of plant functional traits, and help to improve simulations of seasonal carbon and water fluxes in temperate forests.