<![CDATA[Several meteorological practitioners have utilized machine learning techniques to forecast adverse weather conditions, particularly lightning occurrences. Upper air data, obtained through radiosonde measurements, is frequently employed to train machine learning models due to its ability to capture atmospheric instability. Despite its common usage, radiosonde-based lightning predictions typically have a validity window of 6-12 hours. However, cumulonimbus cloud formation in tropical regions, the primary source of lightning, typically lasts between 30 minutes to 1-2 hours per phase, casting doubt on the efficacy of radiosonde data for longer-term predictions. Furthermore, variations in local atmospheric patterns result in non-uniform utilization of radiosonde index parameters across different regions. Understanding the relationship between these parameters and lightning events is crucial for atmospheric thermodynamic analysis and region-specific prediction model development. This study examines the correlation between radiosonde index parameters in the Tanimbar Islands and lightning events from cumulonimbus clouds, utilizing indices such as KI, LI, SI, TT, CAPE, and CIN. Results suggest that index sustainability does not consistently correlate with lightning formation, with differing validity periods observed for 3 and 6 hours ahead. The reason index parameters in the form of SI, KI, and TT are only valid for predicting 3 hours ahead during the months of March-April-May, while only KI maintains validity for both 3 and 6 hours ahead at certain times.]]>
Copyrights © 2024
