Numerical study of performance of two lightning prediction methods based on: Lightning Potential Index (LPI) and electric POTential difference (POT) over Tehran area

The electric POTential difference (POT) and the Lightning Potential Index (LPI) performance in predicting the lightning activity is investigated and the probable relationship between them is examined. These two indices have a similar dependency on microphysical variables such as ice, graupel mixing ratios, and also updraft characteristics within the cloud. Regardless of this similarity, the LPI directly calculated from WRF model is a more favorable parameter for predicting the lightning events in comparison with the POT, which requires an extra package (ELEC) model. Ten years' available data over the Tehran area were reviewed and four thundercloud cases with distinct characteristics (CAPE, time-frequency, intensity) were selected. In order to acquire the associated physical properties, four simulations have been done using the WRF-ELEC model, which is initialized with ERA-Interim data. The assessment conducted within both quantitative and qualitative themes to verify the potential probability of predicting lightning events. In qualitative evaluation framework, the horizontal distribution of LPI and POT were compared to the locations of lightning occurrence detected by WWLLN (The World Wide Lightning Location Network) data as well as the total lightning data obtained from LIS (Lightning Imaging Sensor) in the innermost simulation domain. This evaluation shows that the horizontal patterns of LPI and POT are well consistent with the locations of lightning occurrence. Moreover, in accordance with acquired values of correlation coefficients, it could be inferred that LPI has a better performance in the Number Of Lightning flashes (NOL) prediction than POT. Statistical review of our simulated data (LPI and POT) is performed to study the variation of LPI and POT which resulted in a good correlation between them and therefore, any information about lightning event using LPI values with less computational cost is preferable in the considered domain. Another comparison was carried out based on the calculated correlation coefficient between the simulated NOL and the NOL recorded by WWLLN that does not show any significant correlation between them for none of the cases. Since WWLLN has no observatory in the studied area, inconsistency of the simulated NOL with real atmosphere is justifiable.