Turbulence hazard nowcast products and accident case performance analyses (Formerly Poster 760)

Several automated algorithms have been developed at the National Center for Atmospheric Research to detect, nowcast and forecast turbulence using operational numerical weather prediction (NWP) model and observation data as inputs. These algorithms all make use of artificial intelligence techniques, principally fuzzy logic expert systems and random forest data mining. The algorithms include:

- the NEXRAD Turbulence Detection Algorithm (NTDA), which uses fuzzy logic to process data from Doppler weather radars and detect turbulence within clouds and storms;

- the Diagnose Convectively-Induced Turbulence (DCIT) system, which uses a random forest empirical model to fuse NWP model, satellite, radar and lightning data to infer turbulence associated with convective weather;

- the Graphical Turbulence Guidance (GTG) system, which uses a dynamic fuzzy logic algorithm to combine diagnostics derived from NWP model grids to forecast primarily clear-air turbulence (CAT);

- the Graphical Turbulence Guidance Nowcast system, which uses fuzzy logic to combine NTDA, DCIT, and GTG data to create a comprehensive, rapid-update turbulence nowcast over the conterminous US; and

- the Global Turbulence DSS for Aviation, which extends the GTG system globally.

This paper describes briefly how these algorithms have been developed and tuned, and examines their performance for several turbulence accidents and incidents over both the conterminous US and overseas. These cases demonstrate both the strengths and limitations of the turbulence detection and nowcast products and suggest directions for future research.