Modelling sub-grid wetland in the ORCHIDEE global land surface model: Evaluation against river discharges and remotely sensed data

The quality of the global hydrological simulations performed by land surfacemodels (LSMs) strongly depends on processes that occur at unresolved spatialscales. Approaches such as TOPMODEL have been developed, which allow soilmoisture redistribution within each grid-cell, based upon sub-grid scaletopography. Moreover, the coupling between TOPMODEL and a LSM appears as apotential way to simulate wetland extent dynamic and its sensitivity toclimate, a recently identified research problem for biogeochemicalmodelling, including methane emissions. Global evaluation of thecoupling between TOPMODEL and an LSM is difficult, and prior attempts havebeen indirect, based on the evaluation of the simulated river flow. Thisstudy presents a new way to evaluate this coupling, within the ORCHIDEE LSM,using remote sensing data of inundated areas. Because of differences innature between the satellite derived information -inundation extent -andthe variable diagnosed by TOPMODEL/ORCHIDEE -area at maximum soil watercontent, the evaluation focuses on the spatial distribution of these twoquantities as well as on their temporal variation. Despite some difficultiesin exactly matching observed localized inundated events, we obtain a rathergood agreement in the distribution of these two quantities at a globalscale. Floodplains are not accounted for in the model, and this is a majorlimitation. The difficulty of reproducing the year-to-year variability ofthe observed inundated area (for instance, the decreasing trend by the endof 90s) is also underlined. Classical indirect evaluation based oncomparison between simulated and observed river flow is also performed andunderlines difficulties to simulate river flow after coupling with TOPMODEL.The relationship between inundation and river flow at the basin scale in themodel is analyzed, using both methods (evaluation against remote sensingdata and river flow). Finally, we discuss the potential of the TOPMODEL/LSMcoupling to simulate wetland areas. A major limitation of the coupling forthis purpose is linked to its ability to simulate a global wetland coverageconsistent with the commonly used datasets. However, it seems to be a goodopportunity to account for the wetland areas sensitivity to the climate andthus to simulate its temporal variability.