A trial of ensemble flood simulation experiment based upon TIGGE data with a coupled atmosphere-hydrology model

As the driving force and carrier of Soil-Vegetation-Atmosphere Transfer system (SVAT), the land surface hydrological processes can make great feedbacks to the atmosphere and vice versa. These complicated interactions between the hydrological and atmospheric processes indicate the important effect of water cycle in climate system. However, as one of the most fearful consequences of torrential rain occurred in atmosphere, the short-term flood with varies of forming factors has multiple relations to climate, precipitation, topography and soil. So the hydrological processes should be described more precisely in land surface model (LSM) which then could be coupled to numeric weather prediction model (NWPM) to do the flood forecasting and warning. Moreover, due to the high nonlinear character of atmospheric processes and the simplification of models, the rainfall input from NWPM and model parametric uncertainties would make great influence on the flood simulation. Traditional deterministic forecast would no longer satisfy the operational need, more and more operational forecast adopt ensemble forecast in order to decrease the uncertainties mentioned above. Thus, a watershed hydrological model XXT based on both Xinanjiang model and TOPMODEL was developed by introducing the water table depth Z into the soil moisture storage capacity distribution curve (SMSCC) in Xinanjiang model and then the new SMSCC was integrated with the simple architecture of TOPMODEL. After that, XXT was built and used to instead the simple water balance model (SWB) in Noah LSM. This new Noah LSM was validated in the gauge of Champaign and the results suggested that it improved the description of surface sensible and latent heat flux, as well as the lower water balance residual than the old LSM. Finally this new Noah LSM was coupled with Global/Regional Assimilation Prediction System (GRAPES) developed by China Meteorological Administration (CMA). This trial experiment with the coupled atmosphere-hydrology model was based on the datasets from the THORPEX Interactive Grand Global Ensemble (TIGGE) which included in the Observing system Research and Predictability Experiment (THORPEX) planed by WMO. We select several TIGGE members with different perturbation schemes from CMA, ECMWF and NCEP into the initial and boundary inputs of GRAPES to run the coupled model during a torrential rain event from July 20th to 25th, 2008. Every member was run rolling forward every day during the simulation period with the step of 72 hours. The results of the numeric ensemble experiments show the reasonable distribution of precipitation against the observations. Furthermore, the surface runoff depth simulated by new Noah LSM was much higher than the old LSM, much more consistent with the precipitation. The flood discharges at the outlet of Linyi in Shandong in every 72 hours' time quantum for 6 days described the flood process well. Summary and discussions of this trial of experiment were given at the end.