
Computer production of Kindices: review and comparison of methods
Author(s) -
Menvielle M.,
Papitashvili N.,
Hakkinen L.,
Sucksdorff C.
Publication year - 1995
Publication title -
geophysical journal international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.302
H-Index - 168
eISSN - 1365-246X
pISSN - 0956-540X
DOI - 10.1111/j.1365-246x.1995.tb06895.x
Subject(s) - scaling , automation , algorithm , set (abstract data type) , computer science , statistics , production (economics) , mathematics , data mining , engineering , geometry , programming language , economics , macroeconomics , mechanical engineering
SUMMARY The K indices are to be hand‐scaled from analogue magnetograms following morphological criteria expressed in the so‐called Mayaud rules. All observatories used to follow these rules in the routine hand‐scaling of K indices from analogue magnetograms. As a result of digitalization and automation of observatories, however, observers in charge who are experienced in K scaling from analogue magnetograms are becoming increasingly rare. Therefore the organizations running observatories were forced to find compensating machine methods for the production of K indices. The basic features of the K indices and the Mayaud rules are first reviewed, and then the methods developed for computer derivation of K indices are discussed. In order to decide which of the proposed algorithms was suitable, a comparison among them was organized by the IAGA Working Group on geomagnetic indices. The comparison was made with a common data set, using the same statistical tests. The results are summarized and discussed. Further comparisons between the K indices produced by computer with the different methods are presented in this paper. Four methods were acknowledged by the Working Group during the Vienna IUGG general Assembly in 1991. The results confirm that these methods provide acceptable results in comparison to the average quality of hand‐scaled indices. A comparison between computer‐produced K indices and reference K indices hand‐scaled following the Bartels‐Mayaud rules is also presented. It appears that only two acknowledged methods provide computer‐produced K in good agreement with hand‐scaled K . The best computer method (the FMI method) is found to be good enough to allow the continuation of the long tradition of producing K indices without any serious jump in the statistics.