Premium
Truncation of the EOF series representing 500 mb heights
Author(s) -
Rinne Juhani,
Järvenoja Simo
Publication year - 1979
Publication title -
quarterly journal of the royal meteorological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.744
H-Index - 143
eISSN - 1477-870X
pISSN - 0035-9009
DOI - 10.1002/qj.49710544612
Subject(s) - series (stratigraphy) , empirical orthogonal functions , truncation (statistics) , mathematics , residual , statistics , variance (accounting) , representativeness heuristic , degrees of freedom (physics and chemistry) , econometrics , algorithm , geology , physics , paleontology , accounting , business , quantum mechanics
The representativeness of different sets of EOFs describing 500 mb heights over the northern hemisphere north of 20° is discussed. Some criteria for truncating series given in EOFs are introduced. A case study is presented to find out how well charts constructed with the aid of EOFs represent the original synoptic analysis. In the EOF series, about 160, 130 and 120 terms are needed when applying functions based on analyses from one year, 4 years and 20 years, respectively. Thus the degrees of freedom of the EOF series are cut down to one tenth of those in the corresponding gridpoint presentation (1404 points). From the error variance of the analyses, assumed to be 1.5% of the total sample variance, 0.5 percentage units are included in the EOF series, while the remaining 1.0% is rejected into the residual. The true information lost when truncating the EOF series is 0.5% of the total variance. The EOFs are stable; functions based on data from 1946 work reasonably well in independent data from 1965.