Comparison of measurements with prediction methods for propagation by diffraction at 88-108 MHz

Three methods for predicting attenuation due to diffraction are tested against a large database of 115 614 measurement points, representing 115 614 different path profiles of 100-m horizontal resolution, and vertical root mean square error of about 6 m. The signal level at each geographic measurement point is calculated as the median of about 40 basic measurements, and the measurement database is thus reduced from about four million basic measurements. The mobile measurements are taken from main FM (88-108 MHz) broadcasting emitters located in southern Norway, and the corresponding broadcasting antenna diagrams have been measured by helicopter. Path profiles are categorized by number of terrain obstructions between emitter and receiver, in order to study their effect on each propagation loss method. The current ITU method and Picquenard's construction, with a variable number of included terrain obstructions, are compared with the measurements, and difference statistics are calculated. A particular version of Picquenard's construction is shown to be better than the current ITU method in terrain of Norwegian type. This new method is, in contrast to the ITU method, within the estimated expected errors resulting from using Norwegian digital terrain elevation data.