| Popis: |
Within land surface models (LSM), the biomass allocation scheme (BAS) allows to simulate the dynamics of vegetation growth in response to climatic variation and other drivers. It distributes the assimilated carbon across different biomass pools, and consequently determines the spatio-temporal variability of the leaf area index (LAI).In many LSM, large uncertainties are associated with the BAS, which propagate via the prognostic LAI to the surface fluxes. Here, we propose a revision to the BAS of the ISBA land surface model, by incorporating the dynamics of non-structural carbohydrates (NSC) explicitly. The target of the proposed BAS is the reproduction of LAI as observed with remote sensing, coupled to the modelled surface fluxes. Using in situ eddy covariance observations of the carbon fluxes, and remote sensing observations of the leaf biomass, estimates can be made of the biomass allocation and NSC dynamics. By combining this dataset with other (climatological) variables, a machine-learning based BAS is developed.The simulated evolution of the biomass pools is evaluated using in situ observations of leaf turnover and remote sensing observations of leaf biomass. The proposed model is compared to the standard photosynthesis-driven BAS of ISBA and the more advanced BAS in ORCHIDEE. |
