Some methods for evaluating trends in time series of tropospheric radio field strength data

This paper presents a short review of several methods for evaluating the "seasonal trend" in a series of hourly median field strengths corresponding to one hour of the day. The presentation is from the viewpoint of descriptive statistics, and no discussion is given of the difficult problem of statistical tests of significance for appraising the components of trend found by these methods. Of three methods investigated in CRPL, a type of harmonic analysis was found to be the most successful. The other two methods were an adaptation of the smoothing theory of Wiener and Levinson and a method using orthogonal polynomials. The tools of the statistician, the autocorrelation function and the variate difference method, enable us to refine our analysis of time trends to a point where we hope to obtain a maximum amount of correlation with similar trends in relevant meteorological information.