Improved simulation of carbon and water fluxes by assimilating multi-layer soil temperature and moisture into process-based biogeochemical model

Abstract

Background: Soil temperature and moisture are sensitive indicators in soil organic matter decomposition because they control global carbon and water cycles and their potential feedback to climatic variations. Although the Biome-Biogeochemical Cycles (Biome-BGC) model is broadly applied in simulating forest carbon and water fluxes, its single-layer soil module cannot represent vertical variations in soil moisture. This study introduces the Biome-BGC MuSo model, which is composed of a multi-layer soil module and new modules pertaining to phenology and management for simulations of carbon and water fluxes. Although this model considers soil processes among active layers, estimates of soil-related variables might be biased, leading to inaccurate estimates of carbon and water fluxes. Methods: To improve the estimations of soil-related processes in Biome-BGC MuSo, this study assimilates ground-measured multi-layer daily soil temperature and moisture at the Changbai Mountains forest flux site by using the Ensemble Kalman Filter algorithm. The modeled estimates of water and carbon fluxes were evaluated with measurements using determination coefficient (R2) and root mean square error (RMSE). The differences in the RMSEs from Biome-BGC MuSo and the assimilated Biome-BGC MuSo were calculated (ΔRMSE), and the relationships between ΔRMSE and the climatic and biophysical factors were analyzed. Results: Compared with the original Biome-BGC model, Biome-BGC MuSo improved the simulations of ecosystem respiration (ER), net ecosystem exchange (NEE) and evapotranspiration (ET). Data assimilation of the soil-related variables into Biome-BGC MuSo in real time improved the accuracies of the simulated carbon and water fluxes (ET: R2 =0.81, RMSE=0.70mm·d− 1;ER:R2 =0.85, RMSE=1.97 gC·m− 2·d− 1;NEE:R2 = 0.70, RMSE = 1.16 gC·m− 2·d− 1). Conclusions: This study proved that seasonal simulation of carbon and water fluxes are more accurate when using Biome-BGC MuSo with a multi-layer soil module than using Biome-BGC with a single-layer soil module. Moreover, assimilating the observed soil temperature and moisture data into Biome-BGC MuSo improved the modeled estimates of water and carbon fluxes via calibrated soil-related simulations. The assimilation strategy is applicable to various climatic and biophysical conditions, particularly densely forested areas, and for local or regional simulation.