Assessment of model-based scintillation variance prediction on long-term basis using Italsat satellite measurements

The objective of this paper is to assess the accuracy of model-based statis tical methods for predicting clear-air scintillation amplitude variance fro m ground-based meteorological measurements on long term basis. Six model-ba sed estimation methods are considered and discussed, two of them including also the use of vertically integrated water vapour content as a predictor. They are derived from synthetic data, obtained by applying an electromagnet ic model to a large historical radiosounding dataset in order to simulate t he received scintillation power at microwave and millimeter-wave. The empir ical methods of ITU-R, Karasawa, and Ortgies are also considered for compar ison. The long-term predictions derived from each method are compared with measurements from the Italsat satellite beacons at 18.7, 39.6, and 49.5 GHz , acquired during 1995 at Spine d'Adda (Milan, Italy) site. The method inte rcomparison is carried out by checking the assumed best-fitting probability density function for the log-amplitude fluctuation variance and by applyin g the considered methods to the available ground-based meteorological measu rements. Statistical results in terms of bias, root mean square value and s kewness of the percentage error are discussed in order to understand the po tential and the limits of each model-based prediction method within this ca se study. Copyright (C) 1999 John Wiley & Sons, Ltd.