عنوان مقاله [English]
نویسندگان [English]چکیده [English]
Land surface parameterization schemes estimate the exchanges of momentum, mass and energy between land surface and the atmosphere. Runoff is one of the important components of the land surface water balance. Parameterization of runoff is difficult because of its dependence on rainfall, soil moisture and topography which vary greatly across time and space. A coarse-resolution land surface scheme cannot explicitly model the complexities of runoff generation in the model grid square. Instead, it aims to represent the major processes via sub-grid parameterizations. A popular solution involves the use of probability distribution functions to represent sub-grid variability.
In this paper, the river discharge in three sub-basins of the Karoon river catchment (Farsiat, Harmaleh and Soosan) simulated by the OSU land surface scheme is compared to that simulated by the coupled OSU-SIMTOP (OSU-SIM) model. The OSU land surface scheme parameterizes runoff based on the probability distribution function (pdf) of the soil infiltration, while the coupled OSU-SIM uses the pdf of the topographic characteristics to model runoff. The two models were run off-line using the atmospheric forcing derived from the Weather Research and Forecasting (WRF) model for the 2-month period of 1 December 2005 to 31 January 2006. The simulations were conducted with a 5×5 km grid spacing over a domain having 106×115 grid points along altitude and longitude, respectively, and centered at 50◦E and 32◦N. The models were calibrated during December 2005, and the simulation results for January 2006 were used to intercompare the models and evaluate their simulations against the observed river discharge at the Farsiat, Harmaleh and Soosan hydrometric stations. The results show that, compared to OSU, the OSU-SIM model had higher efficiency, a smaller mean absolute error (MAE) and lower bias in simulating river discharge in all of the three sub-basins. The higher correlation coefficient between the simulated and observed river discharge and closer to 1 normalized standard deviation of the simulated runoff suggest the superiority of OSU-SIM to OSU in all of the three sub-basins for the evaluation period. The lower skill of the OSU in predicting runoff may be attributed to errors of the WRF model in the rainfall prediction, error in the rainfall-runoff relationship of the OSU, or inaccuracy in the surface parameters, especially those related to the pdf of the soil infiltration, used for the simulations. The comparison between the observed river discharge and that simulated by the OSU model shows the error in the initial conditions, especially those initial conditions of surface water and ground water storage, could also be another source of the error in the simulated discharge. Results also suggest that the performance of OSU-SIM is sensitive to the horizontal resolution of the model. Using low-resolution digital elevation data for calculating topographic index decreases the efficiency of the model and increases the mean absolute error of the OSU-SIM land surface scheme.