1988 Volume 40 Issue 6 Pages 645-702
Definitive internal geomagnetic field models are derived for the epochs 1945, 1950, 1955, and 1960. Each model incorporates all data available to us within a five year period centered on the model epoch. For survey data, weighting was determined by examining the spread of residuals for the data, sorted by source and sorted by location, relative to the previous IGRF models. The solution included local biases for the fixed observatories. An a priori model for each epoch was derived by projecting the GSFC (12/83) model, based on Magsat data, backward in time. This projection was accomplished using a spline fit to annual secular-variation models. The spline coefficients were simultaneously fit to all spherical harmonic secular variation coefficients for the 1940 to 1980 time period. This fit included a full covariance analysis. The projected covariances were part of the a priori model for each epoch. An uncertainty model was adopted which included estimates of the effects of crustal and core fields not represented by the model. Differences between model coefficients midway in time between model epochs were compared to estimated coefficient uncertainties. Coefficient differences were within the estimated uncertainties, confirming the uncertainty model. A test for 1945 indicated that a solution without observatory biases was equal to that with such biases, within the expected uncertainties. Differences between biases from year to year are within the bounds expected based on the predicted uncertainties. The resulting models, their secular variation and their expected uncertainties are discussed in some detail.