Different representations of canopy structure-A large source of uncertainty in global land surface modeling

Land surface models generally use a mosaic approach to represent subgrid variation in canopy structure within a given land cell in two ways: one is based on vegetation cover types (VCT) (e.g., biomes or land cover classes); the other uses plant functional types (PFT)-a further breakdown of VCTs. The two representations largely differ in their representations of vegetation composition and canopy geometries, but both are one-dimensional (1D) treatments. To isolate the effects arising from the canopy structures, idealized experiments in a simple three-dimensional (3D) canopy scenario are conducted to analyze the effects of different representations (VCT, PFT and a newly developed 3D representation) on the simulated surface fluxes. The results show that the conventional VCT and PFT representations may have large discrepancies in terms of radiation transfer, resistance and surface flux calculations, whereas the 3D representation usually has values in between or closer to one of the other representations. The discrepancies in surface flux can be attributed to differences in the radiation absorptions and resistance values of different canopy representations. The resistance values of the 3D representation are more similar to those of the VCT case. For shortwave radiation absorption at small sun zenith angles, the 3D representation shows values closer to those of the PFT case but approaches the values of the VCT case for large sun zenith angles. Sensitivity analysis identified such a large source of uncertainty resulting from different representations of canopy structure in land surface modeling, especially for the VCT and PFT representations currently being used. We recommend that proper canopy structures (at least at the crown level), as well as the corresponding physical processes, need to be incorporated in the future development of land surface models.