U. Schmucker, A spherical harmonic analysis of solar daily variations in the years 1964-1965: response estimates and source fields for global induction - I. Methods, GEOPHYS J I, 136(2), 1999, pp. 439-454
This work is based on a time harmonic and subsequent spherical harmonic ana
lysis of daily variations, observed during 24 months at 94 magnetic observa
tories, 18 of them in the southern hemisphere. Observatories in polar or eq
uatorial jet regions are not included. For global induction research this i
s a return to the classical potential method. It requires a dual spherical
harmonic analysis of horizontal and vertical components, and thereby allows
their separation into internal and external parts. The preceding time seri
es analysis is a harmonic analysis of single Greenwich days, with a subsequ
ent phase shift to zero time at local midnight. The selection of days is gu
ided by the degree of magnetic activity, and the emphasis is on the analysi
s of quiet-time daily variations. To take full advantage of extended period
s of quietness, a new measure is introduced, based on a fixed threshold for
the sum of ap indices on the respective day and the adjoining half-days be
fore and after.
The spherical harmonic analysis is carried out with time harmonics from all
observatories except two, but weights are assigned to them to reduce their
hemispherical imbalance. Time harmonics refer either to mean monthly daily
variations or to those on single days. Noting that daily variations depend
primarily on local time, the usual order of summations in spherical harmon
ic expansions is reversed. For each time harmonic, sums are formed over sph
erical terms of the same order m and ascending degree from n = \m\ onwards.
The first partial sum is with m = p for the local-time part of the pth tim
e harmonic, the remaining sums with m = p +/- 1, m = p +/- 2,... for its pa
rt not moving with the speed of the Sun westwards. Up to the fourth harmoni
c, the spectrum of spherical harmonic coefficients is dominated by the seco
nd local-time term with m = p and n = p + 1, except during solstices. For t
he fifth and sixth harmonics, this dominance is lost in all seasons.
The choice of spherical terms to be included has been guided by an eigenval
ue decomposition of the normal equation matrix to ensure a numerically stab
le least-squares solution. No generalized inverse is used in order to allow
a term-by-term determination of the expansion coefficients. In tests, the
total number of terms has been Varied between one and 36. Numerical instabi
lity sets in with a choice of more than 12 terms, notably in the expansion
for the vertical held. With this number of terms, not more than one-half of
the vertical field and about two-thirds of the horizontal held can be acco
unted for by spherical harmonics, in the global average. With a hypothetica
l network of comparable size, but with evenly spaced observing sites, all 3
6 terms could have been included.