Preliminary results of the NCAR Integrated Turbulence Forecasting Algorithm (ITFA) to forecast CAT

Accurate automated one to twelve hour forecasts of turbulence in a three-dimensional airspace are needed by the commercial and general aviation community, but so far, these forecasts have not generally provided acceptably high detection rates and at the same time acceptably low false alarm rates, especially in clear air at upper levels. This is due in part to lack of observations at the (relatively small) atmospheric scales that affect aircraft motion (microscale), and to uncertainties in our knowledge about the relationship between the meteorologically observable (relatively large) scale and the microscale. Until we have a better understanding of the linkage between the large scale and the microscale, we are forced to use semi-empirical techniques to diagnose the large scales for potential turbulence. It is shown that the discrimination capability of these techniques by themselves is in general poor, but a combination performs better. This paper describes an automated turbulence forecasting procedure (the Integrated Turbulence Forecasting Algorithm, ITFA) whereby several turbulence diagnostics are fit to available observations to produce the forecast. Intense verification exercises have been performed over two winter seasons in which probabilities of yes and no detections were determined by comparisons to observations in the form of pilot reports. The sparseness and qualitative nature of this data produces some unavoidable uncertainty in the verification results, however, preliminary results indicate the automated algorithms our competitive with turbulence forecasts produced by knowledgeable human forecasters.