Premium
VARIANCE CONTRIBUTION OF LUNI‐SOLAR (M n AND SOLAR CYCLE (S c ) SIGNALS TO CLIMATE DATA
Author(s) -
CURRIE ROBERT G.
Publication year - 1996
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/(sici)1097-0088(199612)16:12<1343::aid-joc91>3.0.co;2-i
Subject(s) - environmental science , meteorology , variance (accounting) , bandwidth (computing) , climatology , time series , autocorrelation , raw data , atmospheric sciences , statistics , mathematics , geography , computer science , geology , telecommunications , accounting , business
Using an improved numerical procedure the variance contribution of luni‐solar 18ċ6‐year (M n ) and solar cycle 10–11‐year (S c ) signals to 3234 yearly sampled climate records is studied. They comprise 1179 USA air temperature, 1179 USA rainfall, 59 South African rainfall, 39 sea‐level, 697 tree‐ring, and 81 Chinese dryness/wetness time series. On average both signals contribute 25 per cent to total variance in the raw, or original, data. However, on reflection one realizes this is not a realistic comparison because the yearly sampled records have a Nyquist frequency of 0ċ5 cpy , and so contain power for periods greater than 2 years, whereas the bandwidth of the two signals extends from 30 to only 8 years. This realization prompted design of a NL=10 weight low‐pass digital filter, which, convolved with each raw record, filters out power from 8 to 2 years. Repeating analysis on the 3234 filtered time series shows that over a common bandwidth of 30–8 years the signals contribute a mean 74 per cent to the variance in the records. The reason for the dramatic increase in variance contribution from 30–8 years is that a mean 65 per cent of total variance in the raw records is contained in the bandwidth 8–2 years, and so is outside the scope of this paper. Therefore, on decadal and duodecadal time‐scales the spectrum of climate is ‘signal‐like’ rather than ‘noise‐like’, as radically assumed by statisticans and mathematicians the past 70 years.