Secular Fluctuations of Seasonal Precipitation in Lowland California

Orman E. Granger Department of Geography, University of California, Berkeley 94720

Search for other papers by Orman E. Granger in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Seasonal precipitation series from lowland California (1872–1971) are investigated for trends and fluctuations by various statistical techniques including power spectrum. Linear regression and orthogonal polynomials indicate no significant trend over the past 100 years taken as a whole, except at two stations. Overlapping means show shorter period trends, that are oscillatory, consisting of periods of 10–15 and 30 years. There were no statistically significant differences between the whole period and 30-year normals. However, shorter periods of 10–20 years showed significant differences. The 19th century portion of the record was wetter than the 20th century portion at some stations but drier at others. This difference, however, was not statistically significant. The 1941–70 “normal” was found to be abnormal with respect to preceding non-overlapping 30-year periods. Spectral analyses show significant oscillations in winter–spring precipitation with periods of 25–26 months at the majority of coastal stations examined and 30 years at some others. The interrelationship between the quasi-biennial oscillation in California precipitation which results in spells of wet (dry) even (odd) numbered years is discussed in the context of a similar oscillation in other meteorological variables. Similarly, the absence of a statistically significant 5–6 year oscillation in California precipitation contrary to reported oscillations in interrelated ocean-atmosphere systems is noted. Coherence squared and phase differences are used to show that although the various oscillations are spatially coherent, the phase difference in lag time is quite variable. It is concluded that although significant oscillations and spells have existed and should be taken into account in any attempt at long-term foreshadowing, their predictive power is neutralized by apparently as yet unpredictable phase shifts both spatially and temporally.

Abstract

Seasonal precipitation series from lowland California (1872–1971) are investigated for trends and fluctuations by various statistical techniques including power spectrum. Linear regression and orthogonal polynomials indicate no significant trend over the past 100 years taken as a whole, except at two stations. Overlapping means show shorter period trends, that are oscillatory, consisting of periods of 10–15 and 30 years. There were no statistically significant differences between the whole period and 30-year normals. However, shorter periods of 10–20 years showed significant differences. The 19th century portion of the record was wetter than the 20th century portion at some stations but drier at others. This difference, however, was not statistically significant. The 1941–70 “normal” was found to be abnormal with respect to preceding non-overlapping 30-year periods. Spectral analyses show significant oscillations in winter–spring precipitation with periods of 25–26 months at the majority of coastal stations examined and 30 years at some others. The interrelationship between the quasi-biennial oscillation in California precipitation which results in spells of wet (dry) even (odd) numbered years is discussed in the context of a similar oscillation in other meteorological variables. Similarly, the absence of a statistically significant 5–6 year oscillation in California precipitation contrary to reported oscillations in interrelated ocean-atmosphere systems is noted. Coherence squared and phase differences are used to show that although the various oscillations are spatially coherent, the phase difference in lag time is quite variable. It is concluded that although significant oscillations and spells have existed and should be taken into account in any attempt at long-term foreshadowing, their predictive power is neutralized by apparently as yet unpredictable phase shifts both spatially and temporally.

Save