Evaluation of a WRF model multi-physics ensemble forecasting system for simulation of precipitation over central region of Iran

The chaotic nature of the atmosphere, inexact equations of motions, and gaps in specifying the initial state of the atmosphere result in some uncertainties in the weather forecasting. One of the operational methods that has been employed to overcome this difficulty and improve the accuracy of weather forecasts is ensemble forecasting which gains popularity as a technique in numerical weather prediction due to its improved forecast results. There are some different approaches to develope an ensemble forecasting system including using multi-numerical weather prediction models with different numerics, utilizing different physical parametrization schemes, and perturbing initial and boundary conditions. The aim of the present work is to develope a multi-physics ensemble forecasting system for the Weather Research and Forecasting (WRF) model and to evaluate its performance in forecasting of precipitation as a low predictability parameter. To construct the ensemble members, seven microphysics, two planetary boundary layers (PBL) and six cumulus parameterization schemes are employed. These choices are used to generate 84 memebers of the ensemble system. For running the model, a three-nested domain consisting of 36, 12, and 4 km spatial resolution was used to simulate the precipitation for 12 days during 2015-2016 winter in central region of Iran. The one degree FNL data of National Center for Environmental Prediction (NCEP) are utilized as the initial condition for running WRF. The model outputs are evaluated against observations using some standard metrics of forecasting verification including categorical, continuous and probabilistic indices. The observed 24-h accumulated precipitation measured at 132 synoptic stations of national meteorological organization of Iran was used for verification of simulations. In this study, the general performance of the different members of ensemble system is discussed. The members employing MYJ planetary boundary layer scheme, KF and GF cumulus schemes and WSM6, Goddard and New Thompson schemes are found to be appropriate more than the others for accumulated precipitation simulation in the study region. Furthermore, a multi categorical verification using three threshold of 1, 10, and 50 mm/day showed the suitable appraisal of the total performance of ensemble system. Moreover, increasing in ensemble size results in better outputs. However, it seems that expanding the ensemble members might be limited at an optimum number. It means that the skill of ensemble system doesn’t change significantly when the ensemble members are more than the optimum number. In addition, the results show that using a cumulus parametrization scheme for the spatial resolution of the finest domain of this study (i.e., 4 km) is useful to improve the precipitation simulations. It is worth to mention that this finding is in agreement with reports of other studies.