Abstract
Power spectrum techniques are applied to series of daily, weekly, and monthly average temperature and precipitation values, recorded since 1870 at the Woodstock Climatological Benchmark Station in Maryland, in order to gain a reasonable interpretation of the extent and frequency distribution of periodic variations in these data. Analysis procedures are outlined, and the results presented, interpreted, and collated with the results of earlier literature in some detail.
Apparent short-period variations are found whose periods lie near 3 days, between 5 and 7 days, and between about 15 and 25 days; various of them however, are absent from some portions of the data series and also differ somewhat in character with the season of the year.
Significant long-period variations are more prevalent in the temperature series than in the precipitation series. Spectral peaks in temperature, of periods near 2 years and greater than 50 years, both achieve high levels of statistical significance. The 11-year sunspot cycle, and to some extent its second harmonic as well, is suggested in the temperature data. The double (22-year) sunspot cycle and the longer Brückner cycle, however, are almost totally absent. The basis of Abbot's statistical long-range prediction scheme, which utilizes numerous higher harmonics of the double sunspot cycle, is tested against the Woodstock data, and is found in this caw to lack measurable skill above chance.
