146 Using Random Forest Technique to Create Cloud-to-Ground Lightning Probabilities

Thursday, 10 November 2016
Broadway Rooms (Hilton Portland )
Tiffany C. Meyer, CIMMS/Univ. of Oklahoma and NOAA/NSSL, Norman, OK; and K. M. Kuhlman, D. M. Kingfield, and D. J. Gagne II

A new Multi-Radar/Multi-Sensor (MRMS) cloud-to-ground (CG) lightning probability product has been created using machine learning methods. With storm-based inputs of Earth Network’s in-cloud lightning, Vaisala’s cloud-to-ground lightning, MRMS products including the Maximum Expected Size of Hail (MESH) and Vertically Integrated Liquid (VIL), and near storm environmental data including lapse rate and CAPE, a random forest algorithm was trained to produce probabilities of CG lightning.  The algorithm was trained using over 91.000 storms from 36 randomly selected days in 2014. The domain that covered the CONUS. Recently, the algorithm has also been tailored to individual NWS regions to address geographically-variable convective modes.  We will compare the CONUS configuration to the regional variation to determine if a regional product is a necessary direction for real-time implementation of the algorithm.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner