چكيده لاتين :
Studies on rainfall distribution in catchments suffer from scattered distribution and limited number of measuring stations in the catchment area, there is no possibility for detailed analysis. Currently, numerical simulations are the best approach to determine real precipitation on a regular grid in an entire basin. At present, dynamical and physical governing equations of numerical models based on the latest numerical methods provide extensive and valuable information. . Hence, these models are used independently to evaluate occurrence and changes of various atmospheric parameters or effects of changes in one parameter on the others.
In this study, we have evaluated performance of different micro-physical and convection parameterization schemes of WRF model, to estimate precipitation in the Karoon river basin in Southwest Iran. The basin is a main water source of the country, and because of its large water reservoir, the region has been always interesting to researchers. Therefore, the region was selected for study. A mountainous region lies to North and northeast of the basin and the Khuzestan plain in South and West to the area.
Physics of the numerical Weather Research and Forecast (WRF) model, which is developed by different USA institutes, contains 5 categories of: (1) microphysics, (2) cumulus parameterization, (3) planetary boundary layer (PBL), (4) land-surface model, and (5) radiation. Since from its introduction, Operational use of the model has grown significantly. To have a better prediction, it is essential to find and apply the most appropriate configuration for the model in Iran.
The WSM3 is a simple-ice scheme, which predicts three categories of hydrometers: vapor, cloud water/ice, and rain/snow. The scheme computes ice processes efficiently, but not super-cooled water and gradual melting rates. The WSM5 scheme is similar to the WSM3 scheme and includes vapor, rain, snow, cloud ice, and cloud water in five different arrays. Thus, it keeps super-cooled water and gradual melting of snow falling from melting layer. The scheme is efficient in intermediate grids between the meso-scale and cloud-resolving grids.
In practice, the convection parameterization include following steps: Triggering (Determines occurrence/localization of convection), Cloud modeling (Determines vertical distribution of heating, moistening and momentum changes) and Closure (Determines overall amount of the energy conversion, convective precipitation=heat release). Types of convection schemes are based on moisture budgets, Adjustment and Mass-flux schemes. The Kain-Fritsch scheme utilizes a simple cloud model with moist updrafts and downdrafts, including effects of detrainment and entrainment, and a relative simple microphysics. In the Betts-Miller-Janjic scheme, deep convection profiles and the relaxation time are variable depending on the cloud efficiency and a non-dimensional parameter that characterizes the convective regime. The cloud efficiency depends on the entropy change, precipitation, and Cloud mean temperature. A requirement for the shallow convection moisture profile is that the entropy change should be small and non-negative.
Materials and methods
Rainfall prediction is one of the important applications of numerical weather prediction models. Effects of different physical schemas and their combinations were studied in order to select best schemes for more accurate rain prediction, in the very important karoon basin catchment of Iran. To pursue this goal, a matrix of 6 WRF model configurations, were created, using combinations of different microphysical-convection schemes, and were run in two distinct domains with horizontal resolutions of 27km and 9km respectively, for four cases of: January 2004, March 2005 and 2007, and December 2009. In all runs, two different treatments of convection ( i.e., Kain-Fritsch(KF) and Betts-Miller-Janjic (BMJ)) and three different microphysical schemes ( i.e., WSM 3-class(3), WSM 5-class(5) and Ferrier(F)) were used. Also, FNL data from NCEP and observation data from IRIMO were used. Model results were compared with 6-hourly observed data of precipitation from 15 synoptic stations in the region. To evaluate prediction accuracy of different schemes, mean squared correlation coefficients between observation data and each combined convective-microphysics scheme was calculated.
Results and discussion
Mean squared correlation coefficients between observation data and combination of convective-microphysical schemes for 9-km resolution were 0.888 for BMJF, 0.885 for BMJ5, 0.831 for BMJ3, 0.887 for KF3, 0.878 for KF5 and 0.871 for KFF. Therefore, results of the studied cases show no significant difference among the convective-microphysics configurations for the 9km resolution, which are in agreement with the results obtained by Jankov (2005) and Otkin (2008). So it seems that, at intermediate scales (about 9km), regardless of the sensitivity of the model to the microphysics and convection schemes, any of the compound schemes used in this study is acceptable. Therefore, for accurate analysis of schemes due to their differences in physical characteristics, it is better the Model be used in lower scales (for Convective cells). Also, application of preprocessing methods for Observational data assimilation, can detect behavior and sensitivity of the model regarding various schemes. On the other hand, if the convective scale is larger than the grid scale, the model directly resolves convection and precipitation. So, in this case, changing parameterization schemes have little effect on the size and intensity of rainfall. To realize the subject, a control run, without convection Parameterization schemes, should be compared with these model runs.
Conclusion
According to the results obtained the studied cases, there were no significant differences among the diverse convective-microphysics configurations for 9km resolution. Therefore, for an accurate schemes analysis due to differences in their physical characteristics and compare the results, it is better to use the Model in lower scales (higher resolutions) (for Convective cells).
