Summary The near‐Earth magnetic field is caused by sources in the Earth's core, ionosphere, magnetosphere, lithosphere and from coupling currents between the ionosphere and the magnetosphere, and between hemispheres. Traditionally, the main field (low degree internal field) and magnetospheric field have been modelled simultaneously, with fields from other sources being modelled separately. Such a scheme, however, can introduce spurious features, especially when the spatial and temporal scales of the fields overlap. A new model, designated CM3 (Comprehensive Model: phase 3), is the third in a series of efforts to coestimate fields from all of these sources. This model has been derived from quiet‐time Magsat and POGO satellite and observatory hourly means measurements for the period 1960–1985. It represents a significant advance in the treatment of the aforementioned field sources over previous attempts, and includes an accounting for main field influences on the magnetosphere, main field and solar activity influences on the ionosphere, seasonal influences on the coupling currents, a priori characterization of the influence of the ionosphere and the magnetosphere on Earth‐induced fields, and an explicit parametrization and estimation of the lithospheric field. The result is a model that describes well the 591 432 data with 16 594 parameters, implying a data‐to‐parameter ratio of 36, which is larger than several popular field models.

sented by: R n = 1.349 X 109 (0.270) n (nT) 2 for the core and R n = 37.1 (0.974) n (nT) 2 for the crust. These representations can be used to establish order of magnitude inaccuracies in core field models due to crustal fields and due to inability to "observe" the core field wavelengths beyond n = 13 which are totally obscured by the crustal field.

Magsat data has been analyzed as a function of the Dst index to determine the first degree/order spherical harmonic description of the near‐earth external field and its corresponding induced field. The analysis was done separately for data from dawn and dusk with the following results: dusk external: q10 = 20.3 ‐ 0.68 Dst (nT); dusk internal: g10 = 29987.7 + 0.240q10 (nT); dawn external: q10= 18.62 ‐ 0.63 Dst (nT); dawn internal: g10 = 29992.3 + 0.287q10 (nT), where g10 and q10 are the degree 1, zero‐order, internal and external coefficients, respectively, in a spherical harmonic potential function describing the near‐earth magnetic field. Comparison with POGO data indicates that the constant term relating q10 and Dst has changed about 20 nT between 1970 and 1980, presumably due to an increase in the average ring current intensity. A local time variation of the external field persists even during very quiet magnetic conditions. Both a diurnal and 8‐hour period are present. A crude estimate of sq current in the ±45° geomagnetic latitude range is obtained for 1966–1970. The current strength, located in the ionosphere and induced in the earth, is typical of earlier determinations from surface data, although its maximum is displaced in local time from previous results.

A best current model of the main geomagnetic field is presented as a response to a need for an "International Geomagnetic Reference Field ". This model is described by a series of 120 spherical harmonic coefficients and their first and second time derivatives from an epoch 1960.0. It was derived from a sample of all magnetic survey data available from the interval 1900-1964 plus a recent global distribution of preliminary total field observations from the OGO-2 (1965-81A) spacecraft for epoch 1965.8. A duplicate data selection was made and the resulting field model compared with the first to help evaluate the minimum error. It was noted that the root-mean-square difference between the two models was about 30r in the force components, 0.04 degrees in dip and 0.3 degrees in declination at the earth's surface for 1965.0.

Data from the Magsat spacecraft for November 1979 through April 1980 and from 91 magnetic observatories for 1978 through 1982 are used to derive a spherical harmonic model of the earth's main magnetic field and its secular variation at epoch 1980.0. The model is called GSFC(12/83). Constant coefficients are determined through degree and order 13 and secular variation coefficients through degree and order 10. The first‐degree external terms and corresponding induced internal terms are given as a function of Dst. Preliminary modeling using separate data sets at dawn and dusk local time showed that the dusk data contain a substantial field contribution from the equatorial electrojet current. The final data set was therefore selected first from dawn data and then augmented by dusk data to achieve a good geographic data distribution for each of three time periods: (1) November–December 1979, (2) January–February 1980, and (3) March–April 1980. A correction for the effects of the equatorial electrojet was applied to the dusk data utilized. The solution included calculation of fixed biases, or anomalies, for the observatory data. Although similar in many respects, GSFC(12/83) differs from International Geomagnetic Reference Field 1980 by 3.6 nT in the g10 term and shows a slightly negative B in the northern polar region as well as other differences in secular variation pattern.