Impacts from the Processing of Vegetation Types Within a Model Grid Cell

Traditionally, when coupling a land surface model into a GCM, the vegetation types within a model grid cell are grouped as one type, by which some discrepancy of model results to the reality are certainly led to. To better understand the impact, the simulated study through offline experiments of two cases in the same grid cell are conducted. Underlying surface in the cell is considered in one case as the coexistence of multi-vegetation types and the other as one single typical vegetation type.A practical scheme is suggested for this kind of coupling, i.e., a model grid cell is divided into several subgrid cells with each one being called "Tile" according to the number of vegetation types within the grid cell in the form of high resolution data. Meanwhile the land surface model is designed to be able to run simultaneously at different subgrid cells, rather than further to group the vegetation types from the high resolution data. By doing so, the same forcing is shared by all the subgrid cells within one grid cell as in the conventional way, but the ground fluxes need to be integrated from the subgrid cells by percentage of vegetation area coverage. As a matter of fact, the approach can be easily realized. In the meantime, the way to estimate effective ground temperature, effective sensible and latent heats and so on is proposed.In the offline experiments, the land surface model of SSiB and the field observation data of HAPEX-MOBILHY, which are generously provided by Xue who is also the SSiB author, are used to carry out the study. As for real condition, the field of HAPEX-MOBILHY which consists of forest with the percentage of 40% and mixed agricultural crops with the percentage of 60%should be categorized to grass land according to the classification in SSiB. Outputs of simulations show that there are big differences in the effective ground temperature, effective sensible and latent heats predicted respectively by three cases of the land surface including forest, grass and the coexistence of forest and grass. Also, the outputs show that the vegetation type with less percentage cover in a grid cell may make more contribution to the latent or sensible heat in the cell than the vegetation type with more percentage cover in the same grid cell. The validity of the method which considers the vegetation with more percentage in a cell as one representative vegetation type in the cell for some previous land surface model should be reevaluated.