Abstract The impact of initial conditions in predicting lightning activity using the Weather Research and Forecasting (WRF) with the electrification (ELEC) extra package, known as WRF-ELEC model, has been investigated. The severity, frequency, and some physical and dynamical properties of lightning occurred during 11 years (2004–2014) are considered. Four thundercloud events with distinctive characteristics over the Tehran area were chosen for detailed study. The selection process was made based on the observational data received from the Iranian Meteorological Organization (IRIMO), ground-based lightning data from the World Wide Lightning Location Network (WWLLN) and satellite-based lightning data from the Lightning Imaging Sensor (LIS). The WRF-ELEC simulations initialized with the European Centre for Medium-Range Weather Forecasts-ERA Interim (ECMWF-ERA Interim), National Centers for Environmental Prediction Final Analysis (NCEP-FNL) and the NCEP operational Global Forecast System (GFS) data have been conducted to obtain the Number Of Lightning flash density (NOL) associated with each case study. The numerical simulations are compared qualitatively and quantitatively with the observations. Several statistical metrics are used in order to determine the performance of the WRF-ELEC model initialized by various IBCs (initial and boundary conditions) for lightning prediction purposes. The results show that in the most of the studied cases, there is a good agreement between the simulated time-averaged horizontal patterns of the Lightning Potential Index (LPI) obtained from the ERA-Interim-based experiments and the locations of lightning occurrence of WWLLN data as well as LIS observations. Furthermore, GFS-based simulations have a better quantitative performance in the NOL prediction than FNL and ERA-Interim-based simulations regarding the values of Standard Deviation (SD), and centered Root Mean Square Error (RMSE). The results of further statistical analysis using different metrics reveal that ERA-Interim initialization has the best performance of lightning activity prediction.

中文翻译：

---

使用 WRF-ELEC 模型预测闪电活动：初始和边界条件的影响

摘要 已经研究了初始条件对使用天气研究和预报 (WRF) 和电气化 (ELEC) 额外包（称为 WRF-ELEC 模型）预测闪电活动的影响。考虑了 11 年（2004-2014 年）发生的闪电的严重性、频率以及一些物理和动力学特性。选择了德黑兰地区具有鲜明特征的四个雷云事件进行详细研究。选择过程是根据从伊朗气象组织 (IRIMO) 收到的观测数据、来自全球闪电定位网络 (WWLLN) 的地面闪电数据和来自闪电成像传感器 (LIS) 的卫星闪电数据进行的。使用欧洲中期天气预报中心 - ERA Interim (ECMWF-ERA Interim)、国家环境预测最终分析中心 (NCEP-FNL) 和 NCEP 业务全球预报系统 (GFS) 数据初始化的 WRF-ELEC 模拟已经进行以获得与每个案例研究相关的闪电次数密度 (NOL)。数值模拟与观测进行定性和定量比较。使用多个统计指标来确定由各种 IBC（初始和边界条件）为闪电预测目的初始化的 WRF-ELEC 模型的性能。结果表明，在大多数研究案例中，从基于 ERA-Interim 的实验中获得的闪电电位指数 (LPI) 的模拟时间平均水平模式与 WWLLN 数据的闪电发生位置以及 LIS 观测之间存在良好的一致性。此外，在标准偏差 (SD) 和中心均方根误差 (RMSE) 值方面，基于 GFS 的模拟在 NOL 预测中比基于 FNL 和 ERA-Interim 的模拟具有更好的定量性能。使用不同指标的进一步统计分析结果表明，ERA-Interim 初始化具有最佳的闪电活动预测性能。在标准偏差 (SD) 和中心均方根误差 (RMSE) 值方面，基于 GFS 的模拟在 NOL 预测中比基于 FNL 和 ERA-Interim 的模拟具有更好的定量性能。使用不同指标的进一步统计分析结果表明，ERA-Interim 初始化具有最佳的闪电活动预测性能。在标准偏差 (SD) 和中心均方根误差 (RMSE) 值方面，基于 GFS 的模拟在 NOL 预测中比基于 FNL 和 ERA-Interim 的模拟具有更好的定量性能。使用不同指标的进一步统计分析结果表明，ERA-Interim 初始化具有最佳的闪电活动预测性能。