عنوان مقاله [English]
General circulation models (GCMs) provide valuable forecasts of world precipitation and temperature (Schepen et al., 2020). Through improved Seasonal forecasting in recent years several climates centers around the world provide operational climate Such as; the Climate Forecast System version 2 (CFSv2) by National Centers for Environmental Prediction (NCEP) (Saha et al., 2010), the European Centre for Medium-Range Weather Forecasts (ECMWF (Johnson et al., 2019), and the Geophysical Fluid Dynamics Laboratory (GFDL) (Delworth et al., 2020). These GCM outputs generally need to downscale to use in regional-scale relevant applications and more actionable end-user-oriented climate services. One way to transfer world predictions from GCMs to regional or local scales is dynamical downscaling with RCMs such as Weather Research and Forecasting model (WRF). The Initial and lateral boundary conditions from General Circulation Models (GCMs) drive these models. The mesoscale circulations, topography, and land use-land cover are displayed better by RCMs, and these models improve the extremes and regional climate variable compared to the coarse resolution GCMs. The WRF has been coupled with numerous parameterizations to resolve processes occurring within a grid box. Some research has indicated convective and planetary boundary layer (PBL) schemes have a significant influence on precipitation simulation (Li et al 2017; Njuki, S.M., et al 2021). The WRF Model version 4 provides more than 11 convective schemes and 13 planetary boundary layer (PBL) schemes. This study has attempted to assess a suitable combination of physics schemes of the Weather Research and Forecasting (WRF) model for seasonal precipitation simulation over the northeast of Iran. Using the CFSV2 as Initial and lateral boundary conditions data, simulation experiments from winter to spring in seven months (from November to May) have been performed for 2019-2020). Three nested domains have applied with the outer domain at 54 km resolution and two interdomains at 18 and 6 km resolution.
Material and methods:
The study area is located in the northeast of Iran, and climatologically, most precipitation occurs from winter to spring (November to May). On average, the western part of this region receives approximately 60% of the annual precipitation, while the rest of the areas in the east receive lower precipitation. The real-time forecast data used in this study is the 6-hourly time series from the 9-month runs operational model for seasonal prediction at the NCEP operational CFSv2. The observed precipitation data is extracted from IRIMO. The new Weather Research and Forecasting model (WRF) is applied to determine how varying physical parameterization of PBL scheme configuration processes simulate seasonal (winter and spring) precipitation. For this purpose, four group configurations have been designed.
Group1: convective schemes (KFT, BMJ, GF, KF), Yonsei University PBL (YSU) for the plenary boundary layer, surface layer scheme (Revised MM5), the shortwave radiation scheme (Dudhia), the longwave radiation scheme (RRTM) and land surface models (Noah).Group 2 all four convective schemes, PBL Mellor–Yamada–Janjic (MYJ), RRTMG for long –short radiation, 5-layer thermal diffusion, and Eta for land surface and surface layer. GROUP 3; include second-order Mellor-Yamada-Nakanishi-Niino (MYNN3) as PBL scheme, same shortwave and longwave radiation (New Goddard), the surface layer (MYNN), and land surface (RUC). Finally, group4 set by ACM2 for the plenary boundary layer, The surface layer (Pleim-Xiu), the shortwave and longwave radiation schemes (GFDL), the land surface (PX), four convective schemes have been fixed in all groups. For all WRF simulations, we used the WRF single-moment 6-class microphysics scheme. In this way, a total of 20 simulation sets in 4 groups have run, and one configuration set without any cumulus scheme in domain 3 in each group.
The following statistics, the correlation (R), the root mean square error (RMSE), the mean absolute error (MAE), and bias and four verification skills are calculated from the total daily precipitation over the six months out of the seven-month integration time with the first month used as spin-up.
The WRF-CFSv2 model performance was evaluated against precipitation observations from Iran's Meteorological organization. The correlation scores between the observed and predicted 6- month and winter precipitation were moderately acceptable (0.3-0.5) however decreased to 0.36 in spring. In terms of bias, group 1 (PBL1,..) configuration have considerably structures than the group4 (PBL7,..), group2 (PBL2,…), and especially group3 (PBL6,..). All configurations showed a wet bias over the study area (-0.8 mm/d, -3.55mm/d) in the 6-month prediction. It is consistent with previous studies using GCMs in this region. The significant MAE of the 6-month precipitations simulated by group 1 and PBL1-CU2، PBL1-CU0, and PBL1-CU1 scenarios were the lowest among the configuration. Meanwhile, this group of configurations showed a similar RMSE score pattern by MAE, and the lowest RMSE showed in group 1 and group 2. In all configurations, the wet bias has been persistent in the study area.
The WRF prediction captured the observed precipitation by groups 2 and 3 with MYJ and MYNN3 planetary boundary layer schemes. However, the false alarm (b) in group 1 and the number of missed events (c) in group 2 of configurations were finer low.
In this study, the WRF model performance was evaluated for various physical parameterizations in predicting precipitation for varying planetary boundary layer (PBL) schemes and Cumulus schemes over northeast Iran.
Based on the sensitivity analysis, is concluded that the set that performs best for the region is YSU PBL, MM5 SL, Dudhia shortwave radiation, RRTM longwave radiation, and Noah LSM schemes.
And using a cumulus scheme for grid resolutions between 3km and 10km is gray, as respects is not clear whether a cumulus scheme should be used or not. So, recommended testing a configuration set of no cumulus scheme mode to determine if using a cumulus scheme is ideal for your particular run.
Seasonal prediction, CFSRV2, WRF, precipitation