Prediction of Energy Dissipation Rates for Aviation Turbulence. Part I: Forecasting Nonconvective Turbulence

Abstract

urrent automated aviation turbulence forecast algorithms diagnose turbulence from numerical weather prediction (NWP) model output by identifying large values in computed horizontal or vertical spatial gradients of various atmospheric state variables (velocity; temperature) and thresholding these gradients empirically to indicate expected areas of ?light,? ?moderate,? and ?severe? levels of aviation turbulence. This approach is obviously aircraft dependent and cannot accommodate the many different aircraft types that may be in the airspace. Therefore, it is proposed to provide forecasts of an atmospheric turbulence metric: the energy dissipation rate to the one-third power (EDR). A strategy is developed to statistically map automated turbulence forecast diagnostics or groups of diagnostics to EDR. The method assumes a lognormal distribution of EDR and uses climatological peak EDR data from in situ equipped aircraft in conjunction with the distribution of computed diagnostic values. These remapped values can then be combined to provide an ensemble mean EDR that is the final forecast. New mountain-wave-turbulence algorithms are presented, and the lognormal mapping is applied to them as well. The EDR forecasts are compared with aircraft in situ EDR observations and verbal pilot reports (converted to EDR) to obtain statistical performance metrics of the individual diagnostics and the ensemble mean. It is shown by one common performance metric, the area under the relative operating characteristics curve, that the ensemble mean provides better performance than forecasts from individual model diagnostics at all altitudes (low, mid-, and upper levels) and for two input NWP models.