Multi-parameter approach for the detection of lightning activity using a low-cost and portable X-band radar

This work proposes a multi-parameter method for the detection of cloud-to-ground (CG) lightning activity associated to convective cells, based on the measurements performed by a low-cost and portable single-polarization X-band weather radar. With respect to previous studies on the same topic, in this work various lightning proxy parameters are extracted for each rain cell observed by the radar and properly merged to produce an estimation of the expected lightning stroke activity for each rain cell. The developed methodology is based on a modified Storm Cell and Identification Tracking method for the convective cell identification and on a quadratic discriminant analysis for the combination of the lightning proxy parameters into a single label, indicating two categories of CG activity: flash rate equal to zero or greater than zero, respectively. Three lightning radar based proxy parameters have been selected after a correlative analysis, which considered a large number of predictors. The three radar based lightning proxy parameters selected are: i) the altitude thickness between the iso-reflectivity at Z=40-dBZ and the T=-20°C isotherm; ii) the rain cell area contoured by iso-reflectivity at Z=40-dBZ found at T=-10°C isotherm and iii) the vertical integral of ice equivalent water. In order to train and test our procedure, a dataset collected in the pilot study area of Naples metropolitan environment and consisting of 1575 radar reflectivity volumes acquired from April 2012 to March 2016, LINET lighting network of CG and intra-cloud (IC) strokes and meteorological in-situ data has been used. Radar data have been processed, in order to remove clutter and partially compensate for attenuation effects. The results, discussed in terms of detection scores, show that the developed technique, although with some limitations, outperforms those that make use of single lightning proxy parameters. Overall, the main results are fairly reasonable compared with other studies and more advanced radar systems, showing a Proportion Correct (PC) detection of 0.60, a Probability of Detection (POD) of 0.67, a False Alarm Ratio (FAR) of 0.29, a Critical Success Index (CSI) of 0.52 and a Heidke Skill Score of 0.47.