3:30 PM - 3:45 PM
[ACG39-07] Data assimilation of multilayer SIF emission by VISIT-SIF for better GPP estimation in a cool-temperate forest in Takayama
Keywords:Terrestrial ecosystem model, Solar-induced chlorophyll fluorescence, Parameter optimization, Data assimilation, Gross primary production, Understory
The terrestrial ecosystem assimilates atmospheric carbon dioxide through photosynthesis which is one of the most important carbon sinks. Solar-induced chlorophyll fluorescence has been well-known as a proxy of photosynthesis and shows a strong correlation with gross primary production (GPP) at the ecosystem level. In recent years, satellite-observed SIF has been applied to optimize the parameters of terrestrial ecosystem models for better carbon cycle simulations.
Takayama research forest (denoted as TKY: 36.13° N, 137.42° E, 1420 m a.s.l.) is the only site observing the multilayer SIF emissions to understand the photosynthetic activity in both overstory and understory. This study focused on improving GPP simulations of overstory and understory using vertical profiled SIF in TKY. We simulated SIF in the TKY from 1979 to 2020 with a half-hourly time step using the VISIT-SIF model which was developed to link SIF and photosynthetic processes. In growing season, incoming radiation was declined through the canopy due to dense growth of the overstory, resulting in restriction of simulated SIF emissions from the understory. This phenomenon was also confirmed in the observation, however, VISIT-SIF underestimated the intensity of SIF emission in the understory. To improve the discrepancy of SIF emissions, we modified model parameters and expected better GPP estimation by updated parameters. The parameters related to phenology and maximum carboxylation rate were optimized by using the Bayesian optimization method to minimize the error between simulated and observed SIF using only top-of-canopy SIF as a preliminary experiment. The results showed substantial improvement in the simulated SIF using optimized parameters, RMSE of 10:00-14:00 averaged SIF from 2019 to 2020 decreased from 0.61 to 0.36 mW m-2 sr-1 nm-1, and RMSE of GPP decreased from 0.10 to 0.08 gC m-2 half hour-1. In our presentation, we will introduce the parameter optimization for GPP simulations of overstory and understory in TKY.
Takayama research forest (denoted as TKY: 36.13° N, 137.42° E, 1420 m a.s.l.) is the only site observing the multilayer SIF emissions to understand the photosynthetic activity in both overstory and understory. This study focused on improving GPP simulations of overstory and understory using vertical profiled SIF in TKY. We simulated SIF in the TKY from 1979 to 2020 with a half-hourly time step using the VISIT-SIF model which was developed to link SIF and photosynthetic processes. In growing season, incoming radiation was declined through the canopy due to dense growth of the overstory, resulting in restriction of simulated SIF emissions from the understory. This phenomenon was also confirmed in the observation, however, VISIT-SIF underestimated the intensity of SIF emission in the understory. To improve the discrepancy of SIF emissions, we modified model parameters and expected better GPP estimation by updated parameters. The parameters related to phenology and maximum carboxylation rate were optimized by using the Bayesian optimization method to minimize the error between simulated and observed SIF using only top-of-canopy SIF as a preliminary experiment. The results showed substantial improvement in the simulated SIF using optimized parameters, RMSE of 10:00-14:00 averaged SIF from 2019 to 2020 decreased from 0.61 to 0.36 mW m-2 sr-1 nm-1, and RMSE of GPP decreased from 0.10 to 0.08 gC m-2 half hour-1. In our presentation, we will introduce the parameter optimization for GPP simulations of overstory and understory in TKY.