Outstanding problems for the probabilistic prediction of the performance of HF terrestrial services

Procedures for the long-term prediction of performance of terrestrial HF systems have been based on the prediction of monthly median conditions, with the subsequent application of statistical distributions, or geophysical or solar parameters, for other time percentages. However, there may be little correlation with time or location of ionospheric variability for various ionospheric parameters. The paper will address the issues primarily for the computer program ITU HFPROP (previously called REC533), but it is believed that the issues will also be relevant for other widely used programs.