Forecasting Lightning Cessation Using Data from a Network of Field Mills at Kennedy Space Center and Cape Canaveral Air Force Station

Improving lightning cessation forecasts is important to operations of the 45th Weather Squadron (45WS) at Kennedy Space Center (KSC). If lightning advisories can be cancelled closer to the time that cessation actually occurs without compromising safety, there is the opportunity to save time, money, and improve the scheduling of space launches. This paper describes the use of data from a network of 31 field mill (FM) sensors located at KSC and Cape Canaveral Air Force Station (CCAFS) to determine whether a storm's last flash has occurred based on surface electric field values under thunderstorms. Along with the FM data, radar derived products are utilized in developing and analyzing the forecasting schemes. The dataset consists of warm season, quasi-isolated thunderstorms in east central Florida from 2009 - 2012. Radar products are used from the Tampa (KTBW) and Melbourne (KMLB) National Weather Service radars. Lightning data from the Lightning Detection and Ranging - Second Generation (LDAR-II) network also are utilized. LDAR-II detects source emissions from both intra-cloud (IC) and cloud-to-ground (CG) flashes. Rapid Update Cycle (RUC) and Rapid Refresh (RAP) model analyses are used to describe environmental conditions of the storms. The storms are tracked using the Warning Decision Support System - Integrated Information software (WDSS-II). Based on multiple FM derived variables and maximum reflectivity at -10°C altitude, algorithms are developed using various thresholds and wait times. Then, combinations of FM and radar parameters are used to create new algorithms. Skill scores are calculated for each algorithm based on the number of hits (correct forecasts), false alarms (cancelling an advisory prior to cessation), and misses (never ending an advisory). Since false alarms present the most dangerous situations, they are the most heavily weighted in selecting the best performing algorithms. The times after successful cancellations are determined and compared to current 45WS' approaches. Results indicate that while algorithms using only FM data show forecast skill and save time, they are not sufficiently safe. Conversely reflectivity algorithms produce zero false alarms, but yield too many misses. Algorithms combining the FM and radar data give the best results. The overall best performing algorithm includes the one minute standard deviations of the data from the FM sensor closest to the last flash initiation point (S1CL) at a threshold of 380 V m-1 s-1, a reflectivity threshold of 35 dBZ, and a 5 min wait time. This combination does not prematurely suspend an advisory for any of the storms analyzed. On average, it ends advisories 9.91 min after cessation occurs, an improvement of 20.09 min based on the 45WS' most conservative approach of 30 min. These results show that the FM network at KSC/CCAFS in combination with radar products can be useful in improving lightning cessation forecast guidance.