ارزيابي دبي شبيه سازي شده رودخانه توسط طرحواره سطح NOAH جفت‌شده در مدل WRF با وارد كردن تاثير شاخص توپوگرافي و مدل آب زيرزميني

Evaluation of simulated river discharge by NOAH land surface scheme coupled in WRF by entering the effect of topographic index and groundwater model

رواناب از مهم ترين مولفه هاي طرحواره هاي سطح است كه برآورد آن به دليل وابستگي به بارندگي، رطوبت خاك و توپوگرافي، كه به شدت با زمان و مكان متغير هستند، مشكل است. در اين بررسي دو روش مختلف پارامتره سازي زيرشبكه اي رواناب در طرحواره سطح NOAH جفت شده در مدل پيش بيني عددي WRF در سه زيرحوضه رودخانه كارون (زيرحوضه هاي سوسن، حرمله و فارسيات) در زمستان 2006 مقايسه مي­شوند. رواناب سطحي در طرحواره NOAH براساس تابع توزيع احتمال بيشينه ظرفيت نفوذ خاك و رواناب زيرسطحي براساس نفوذ گرانشي كف مدل خاك پارامتره مي شود. مدل WRF-NOAH با مدل رونديابي رودخانه TRIP در حوضه رودخانه كارون، دبي را به شدت فروبرآورد مي كند كه اين  مي تواند ناشي از عدم قطعيت در پارامتره سازي رواناب سطحي، به دليل در دسترس نبودن داده هاي نفوذ خاك و بكار بردن تابع ذهني در برآورد تابع چگالي احتمال آن، و رواناب زيرسطحي، به دليل برآورد نادرست رطوبت لايه كف مدل خاك، باشد. لذا در اين مقاله طرحواره جديد NOAH-SIM، بر مبناي روش تاپ مدل ساده شده ارائه مي شود كه رواناب سطحي بر اساس شاخص توپوگرافي، به دليل در دسترس بودن داده هاي شاخص توپوگرافي، رواناب زيرسطحي با واردكردن عمق ايستابي و مدل آب زيرزميني پارامتره مي شود. ارزيابي دبي شبيه سازي شده توسط طرحواره هاي سطح NOAH و NOAH-SIM به طور جفت شده در مدل WRF در حوضه رودخانه كارون، بيانگر ضريب كارايي بالاتر، اريبي پايين تر، خطاهاي مدل كوچك تر، ضريب همبستگي بالاتر و انحراف معيار نرمال شده نزديك به يك طرحواره SIM-NOAH است كه برتري پارامتره سازي رواناب بر اساس تاپ مدل ساده شده، به ويژه رواناب زيرسطحي به دليلي حاكم بودن آن در اكثر زمان هاي مورد مطالعه، را نشان مي دهد. هم چنين تاثير پارامتره سازي رواناب بر توازن بودجه آبي در هر دو طرحواره سطح،  بيانگر توازن بهتر بودجه آبي توسط طرحواره NOAH-SIM است.

Introduction Land Surface parameterization Schemes (LSS) play an important role in both general circulation models and regional weather prediction models. A Land surface scheme generally solves the surface energy balance equation to compute the partitioning of the available surface net radiation into sensible and latent heat fluxes, and the surface water balance equation to compute the partitioning of precipitation into evaporation, runoff and the change in the soil moisture storage. Runoff is an important component of the water cycle whose estimation is very difficult because of its strong spatial and temporal variability. In land surface schemes, runoff is usually represented as the sum of the surface runoff and subsurface runoff. Surface runoff occurs due to saturation excess (Dunn Mechanism) and infiltration excess (Horton Mechanism) flows. Subsurface runoff occurs when water enters into the soil and reaches to an impermeable layer or a layer with low permeability. Then, the water flows along the slope under the surface. A coarseresolution land surface scheme cannot explicitly model the complexities of runoff generation in the model grid square. Instead, it represents the major processes via subgrid scale parameterizations. A popular solution involves the use of probability distribution functions (pdf) to represent subgrid scale variability.   Materials and Methods In this paper, two types of runoff parameterizations in NOAH land surface scheme coupled in the Weather and Research Forecasting model (WRF) are examined. The default WRFNOAH parameterizes surface runoff based on the pdf of soil infiltration and subsurface runoff based on the gravity drainage from the lowest layer of the soil model. We have modified the runoff parameterization of NOAH by following the philosophy used in the simplified TOPMODEL, in which surface runoff is parameterized using the pdf of the topographic index and subsurface runoff is defined by applying a simple groundwater model to the lowest layer of the soil model. To obtain low resolution topographic index, the downscaling method of Pradhan et al. is used. To calculate the river discharge the Total Runoff Integrating Pathway (TRIP) is coupled with the land surface scheme. In this study, only the treatment of runoff in the model is considered, hence some of the errors in simulations can be the result of deficiencies in the parameterization of other process, such as precipitation. In this paper, the Karoon River is divided into three subbasins including Farsiat, Harmaleh and Soosan located in the south, west and east of the Karoon respectively by using ARCGIS and ARCHYDRO softwares. The WRF model was run in a oneway method, consisted of two domains. The simulations are conducted for the winter 2006 with 5×5 km grid spacing over an internal domain having 108×114 grid points along latitude and longitude, respectively, and with 15×15 km grid spacing over the parent domain having 69×69 grid points along latitude and longitude and centered at 50◦E and 32◦N. The initial and boundary conditions are derived from the GFS data. The river discharge calculated from WRFNOAH and WRFNOAHSIM simulated runoff and routed using TRIP model for three subbasins of Karoon River is compared with the observed discharge in the winter 2006.   Results and discussion The comparison between the simulated discharge of WRFNOAH and observed discharge shows that the model generally underestimates total runoff during winter 2006, and that there is large model bias and Mean Absolute Error (MAE) in all the three subbasins, particularly in Farsiat and Harmaleh. This is due to the great differences between the mean discharge of the coupled model and what observed. The negative efficiency of the model at Harmale shows that NOAH is not successful in simulating runoff; however, the efficiency of the model is positive, but small in the other subbasins. The daily simulated runoff shows that the modeled peaks that occur due to precipitation were generally too low, and there was little flow during the recession in the studied subbasins. The evaluation of stimulated discharge by the two land surface schemes (NOAHSIM, NOAH) coupled in WRF, with observed discharge proves improved runoff simulation by NOAHSIM in all the three subbasins in the winter 2006. Compared to NOAH, NOAHSIM simulated discharge has lower bias, smaller mean absolute error, higher model efficiency, higher correlation coefficient and the standard deviation closer to that observed. The daily study of simulated and observed discharge shows that NOAHSIM improves surface runoff parameterization by decreasing the differences from observed peaks of discharge. As the subsurface runoff dominates in most of the study period, improved subsurface runoff parameterization of NOHASIM has greatly reduced errors (bias and MAE) in estimating total runoff. The comparison of water balance in these land surface schemes coupled in WRF shows that NOAHSIM has been able to balance the water budget better than what NOAH does in winter 2006.   Conclusion  The results of the comparison between two types of subgrid scale surface runoff parameterization in NOAH land surface scheme coupled in WRF represent the superiority of runoff parameterization based on the topographic index to that based on soil infiltration, probably because of the availability of topographic data and the use of objective function for estimating its pdf. The subsurface runoff parameterization by using groundwater model, instead of gravity drainage, has a great impact on the improvement of total runoff simulation. Lower uncertainties in runoff parameterization of NOAHSIM not only lead to improve discharge simulation but also lead to better water budget balance.