Verification of aviation turbulence detection, nowcast, and forecast products

While pilot reports (PIREPs) have long been the standard for verifying turbulence forecasts, their subjective nature and poor temporal and spatial accuracy severely limit their value in verifying the much more localized turbulence typically associated with clouds and storms. Moreover, the sporadic, conditional nature of their reporting makes them essentially useless for verifying probabilistic forecasts, since they provide a highly biased distribution of turbulence values. Automated, quantitative reports of turbulence from some commercial aircraft help to ameliorate this problem, but they too have limitations such as preferential sampling of the atmosphere in regions thought by pilots to be safe. This paper discusses these challenges in detail and proposes a methodology for using in-situ EDR data for the verification of turbulence products being developed for NextGen, including probabilistic Graphical Turbulence Guidance forecasts and convectively-induced turbulence diagnoses. Sample verification results for these products are also presented.