Evaluating the sensitivity of RegCM4 model to types of Convection Parameterization Schemes on the modeling of springtime precipitation in the North West of Iran: (Case Study: Spring 2004)

In order to implement Limited Area Models (LAMs) for a new area, an appropriate domain size, an ideal horizontal and vertical resolution, an appropriate Initial and lateral boundary conditions, and an appropriate Convection Parameterization Schemes (CPSs) are the most important challenges that should be considered. RegCM4 model provides several options, in relation to convection and land surface schemes. Therefore, one of the very important issues is the choice of an appropriate CPS in the model. In fact, precipitation is the most challenging variable which should be considered in numerical modeling. In this regard, using different CPSs could have a significant impact on precipitation characteristics such as intensity, frequency, and spatiotemporal variations. Various studies have been done to evaluate the sensitivity of convective schemes on the simulated variables, particularly precipitation. Suffice to say that all of the studies reveal the importance and impact of CPSs on the simulation results.
In this study, the Regional Climate Model version 4 (RegCM4.1) was used to evaluate types of CPSs to simulation of springtime precipitation in the North West of Iran (NWI). For this purpose, all conditions such as domain size, vertical and horizontal resolution, initial and lateral boundary conditions have been set uniformly and three experiments were conducted with three CPSs with a 20km horizontal resolution for the Spring 2004 (March, April, May and June) as a typical Spring season. One-month spin-up was considered for the simulation and then at each simulation the model was run at the beginning of February 2004. In order to analyze the effects of the boundary conditions (BCs) on the model simulations, the ERA-interim data with a horizontal resolution of 1.5°´1.5° were employed to provide the lateral boundary conditions for RegCM4. Various types of data were used to evaluation the RegCM4 performance for large-scale and convective precipitation. The monthly and seasonal precipitation of APHRODITE, AgMMERA, CHIRPS and PERSIANN CDR gridded data were used for validation of large-scale precipitation. ERA-Interim data with a horizontal resolution of 0.25°´0.25° spanning 4 months from March to June 2004 were used for validation and evaluation of convective precipitation over two different domains, i.e. the Middle East (ME) and NWI. For evaluation of most appropriate CPS, quantity correlation coefficient, standard deviation, and root mean square error (RMSE) has been demonstrated using Taylor diagrams, and also the bias or average error of precipitation estimation has been calculated
The results show that overall precipitation patterns demonstrated by four Satellite / Rain gauge based gridded precipitation data are coincided to each other. The maximum amount of precipitation in the area has occurred between the Caspian Sea and the Black Sea in accordance with the North Caucasus. The highest amount of seasonal rainfall by the CHIRPS data is demonstrated in the eastern edge of the Black Sea and the southern coast of the Caspian Sea and the lowest amount in the same regions are related to the APHRODITE data. In terms of spatial distribution, AgMMERA and CHIRPS show more details of precipitation occurrence in compared to PERSIANN CDR and APHRODITE data is estimating total precipitation of four months with three Convective schemes, it becomes clear that, there is a high relative agreement between simulated and observed precipitation, in terms of spatial distribution. The highest estimates of total precipitation are related to Kuo scheme which occurs in a small region of the Black Sea coast and the total amount is up to 1600 mm. Evaluation of CPSs in the estimating of convective precipitation has revealed that they have different simulations on the smaller domain of NWI than to larger domain (i.e. ME). Generally, Emanuel scheme has simulated the large-scale precipitation with highest overestimates while the Kuo Scheme has simulated more balanced conditions of the precipitation. Calculation of the error and bias for convective precipitation amounts revealed that all three schemes had a negative bias in simulating of convective precipitation both for ME and NWI. Generally, the model has simulated an underestimation of convective precipitation in comparison with observations. Results also indicate that the RegCM4 model is not very sensitive to the type of CPS to simulate large scale and convective precipitation in the two first months of Spring (i.e. March and April), while the model is more sensitive to simulate the precipitation for May and June. Evaluating the CPSs performance in the simulation of large-scale and convective precipitation indicates that the Kuo scheme has a relative superiority in comparison to Emanuel and Grell schemes. Also, the model simulation with Kuo scheme over the larger domain (i.e. ME) and Emanuel scheme in the smaller domain (i.e. NWI), have the least bias in simulating the total spring convective precipitation. Due to the relatively higher performance of the Kuo Scheme compared to others, this scheme was chosen as the most appropriate way to simulate the large scale and convective precipitation of the Spring over the study area.