Export 16 results:
Sort by: Author Title Type [ Year  (Desc)]
Davies, CJ, Constable CG.  2018.  Searching for geomagnetic spikes in numerical dynamo simulations. Earth and Planetary Science Letters. 504:72-83.   10.1016/j.epsl.2018.09.037   AbstractWebsite

We use numerical dynamo simulations to investigate rapid changes in geomagnetic field intensity. The work is motivated by paleomagnetic observations of 'geomagnetic spikes', events where the field intensity rose and then fell by a factor of 2-3 over decadal timescales and a confined spatial region. No comparable events have been found in the historical record and so geomagnetic spikes may contain new and important information regarding the operation of the geodynamo. However, they are also controversial because uncertainties and resolution limitations in the available data hinder efforts to define their spatiotemporal characteristics. This has led to debate over whether such extreme events can originate in Earth's liquid core. Geodynamo simulations produce high spatio-temporal resolution intensity information, but must be interpreted with care since they cannot yet run at the conditions of Earth's liquid core. We employ reversing and non-reversing geodynamo simulations run at different physical conditions and consider various methods of scaling the results to allow comparison with Earth. In each simulation we search for 'extremal events', defined as the maximum intensity difference between consecutive time points, at each location on a 2 degrees latitude-longitude grid at Earth's surface, thereby making no assumptions regarding the spatio-temporal character of the event. Extremal events display spike-shaped time-series in some simulations, though they can often be asymmetric about the peak intensity. Maximum rates of change reach 0.75 mu Tyr(-1) in several simulations, the lower end of estimates for spikes, suggesting that such events can originate from the core. The fastest changes generally occur at latitudes > 50 degrees, which could be used to guide future data acquisitions. Extremal events in the simulations arise from rapid intensification of flux patches as they migrate across the core surface, rather than emergence of flux from within the core. The prospect of observing more spikes in the paleomagnetic record appears contingent on finding samples at the right location and time to sample this particular phase of flux patch evolution. (C) 2018 Published by Elsevier B.V.

Cromwell, G, Johnson CL, Tauxe L, Constable CG, Jarboe NA.  2018.  PSV10: A global data set for 0-10 Ma time-averaged field and paleosecular variation studies. Geochemistry Geophysics Geosystems. 19:1533-1558.   10.1002/2017gc007318   AbstractWebsite

Globally distributed paleomagnetic data from discrete volcanic sites have previously been used for statistical studies of paleosecular variation and the structure of the time-averaged field. We present a new data compilation, PSV10, selected from high-quality paleodirections recorded over the past 10 Ma and comprising 2,401 sites from 81 studies. We require the use of modern laboratory and processing methods, a minimum of four samples per site, and within-site Fisher precision parameter, k(w), 50. Studies that identify significant tectonic effects or explicitly target transitional field states are excluded, thereby reducing oversampling of transitional time intervals. Additionally, we apply two approaches using geological evidence to minimize effects of short-term serial correlation. PSV10 is suitable for use in new global geomagnetic and paleomagnetic studies as it has greatly improved spatial coverage of sites, especially at equatorial and high latitudes. VGP dispersion is latitudinally dependent, with substantially higher values in the Southern Hemisphere than at corresponding northern latitudes when no VGP cutoff is imposed. Average inclination anomalies for 10 degrees latitude bins range from about +32 degrees to -7.52 degrees for the entire data set, with the largest negative values occurring at equatorial and mid-northern latitudes. New 0-5 Ma TAF models (LN3 and LN3-SC) based on selections of normal polarity data from PSV10 indicate a Non-zonal variations in field structure are observed near the magnetic equator and in regions of increased radial flux at high latitudes over the Americas, the Indian Ocean, and Asia.

Davies, C, Constable C.  2017.  Geomagnetic spikes on the core-mantle boundary. Nature Communications. 8   10.1038/ncomms15593   AbstractWebsite

Extreme variations of Earth's magnetic field occurred in the Levant region around 1000 BC, when the field intensity rapidly rose and fell by a factor of 2. No coherent link currently exists between this intensity spike and the global field produced by the core geodynamo. Here we show that the Levantine spike must span >60 degrees longitude at Earth's surface if it originates from the core-mantle boundary (CMB). Several low intensity data are incompatible with this geometric bound, though age uncertainties suggest these data could have sampled the field before the spike emerged. Models that best satisfy energetic and geometric constraints produce CMB spikes 8-22 degrees wide, peaking at O(100) mT. We suggest that the Levantine spike reflects an intense CMB flux patch that grew in place before migrating northwest, contributing to growth of the dipole field. Estimates of Ohmic heating suggest that diffusive processes likely govern the ultimate decay of geomagnetic spikes.

Cromwell, G, Constable CG, Staudigel H, Tauxe L, Gans P.  2013.  Revised and updated paleomagnetic results from Costa Rica. Geochemistry Geophysics Geosystems. 14:3379-3388.   10.1002/ggge.20199   AbstractWebsite

Paleomagnetic results from globally distributed lava flows have been collected and analyzed under the time-averaged field initiative (TAFI), a multi-institutional collaboration started in 1996 and designed to improve the geographic and temporal coverage of the 0-5 Ma paleomagnetic database for studying both the time-averaged field and its very long-term secular variations. Paleomagnetic samples were collected from 35 volcanic units, either lava flows or ignimbrites, in Costa Rica in December 1998 and February 2000 from the Cordilleras Central and Guanacaste, the underlying Canas, Liberia and Bagaces formations and from Volcano Arenal. Age estimates range from approximately 40 ka to slightly over 6 Ma. Although initial results from these sites were used in a global synthesis of TAFI data by Johnson et al. (2008), a full description of methodology was not presented. This paper documents the definitive collection of results comprising 28 paleomagnetic directions (24 normal, 4 reversed), with enhanced precision and new geological interpretations, adding two paleointensity estimates and 19 correlated Ar-40/Ar-39 radiometric ages. The average field direction is consistent with that of a geocentric axial dipole and dispersion of virtual geomagnetic poles (17.34.6 degrees) is in general agreement with predictions from several statistical paleosecular variation models. Paleointensity estimates from two sites give an average field strength of 26.3 T and a virtual axial dipole moment of 65 ZAm(2). The definitive results provide a useful augmentation of the global database for the longer term goal of developing new statistical descriptions of paleomagnetic field behavior.

Korte, M, Constable C.  2011.  Improving geomagnetic field reconstructions for 0-3 ka. Physics of the Earth and Planetary Interiors. 188:247-259.   10.1016/j.pepi.2011.06.017   AbstractWebsite

Global geomagnetic field reconstructions on millennial time scales can be based on comprehensive paleomagnetic data compilations but, especially for older data, these still suffer from limitations in data quality and age controls as well as poor temporal and spatial coverage. Here we present updated global models for the time interval 0-3 ka where additions to the data basis mainly impact the South-East Asian, Alaskan, and Siberian regions. We summarize recent progress in millennial scale modelling, documenting the cumulative results from incremental modifications to the standard algorithms used to produce regularized time-varying spherical harmonic models spanning 1000 BC to 1990 AD: from 1590 to 1990 AD gauss coefficients from the historical gufm1 model supplement the paleomagnetic information; in addition to absolute paleointensities, calibrated relative paleointensity data from sediments are now routinely included; iterative data rejection and recalibration of relative intensity records from sediments ensure stable results; bootstrap experiments to generate uncertainty estimates for the model take account of uncertainties in both age and magnetic elements and additionally assess the impact of sampling in both time and space. Based on averaged results from bootstrap experiments, taking account of data and age uncertainties, we distinguish more conservative model estimates CALS3k.nb representing robust field structure at the core-mantle boundary from relatively high resolution models CALS3k.n for model versions n = 3 and 4. We assess the impact of newly available data and modifications to the modelling method by comparing the previous CALS3k.3, the new CALS3k.4, and the conservative new model, CALS3k.4b. We conclude that with presently available data it is not feasible to produce a model that is equally suitable for relatively high-resolution field predictions at Earth's surface and robust reconstruction of field evolution, avoiding spurious structure, at the core-mantle boundary (CMB). We presently consider CALS3k.4 the best high resolution model and recommend the more conservative lower resolution version for studies of field evolution at the CMB. (C) 2011 Elsevier B.V. All rights reserved.

Constable, CG.  2011.  Modelling the geomagnetic field from syntheses of paleomagnetic data. Physics of the Earth and Planetary Interiors. 187:109-117.   10.1016/j.pepi.2011.05.004   AbstractWebsite

This review examines results from time-varying geomagnetic field models that span several thousand years, and from variations in dipole moment strength up to million year time scales. For the past 400 years, twin magnetic flux lobes bordering the inner core tangent cylinder in both northern and southern hemispheres dominate the geomagnetic field and appear more or less fixed in location. In contrast, the millennial scale view shows that such features are quite mobile and subject to morphological changes on time scales of a few centuries to a thousand years, possibly reflecting large scale reorganization of core flow. The lobes rarely venture into the Pacific hemisphere, and average fields over various time scales generally reveal two or three sets of lobes, of diminished amplitude. Thus millennial scale models are suggestive of thermal core-mantle coupling generating a weak bias in the average field rather than a strong inhibition of large scale field changes. The recovery of variations in dipole moment on million year time scales allows frequency domain analyses to search for characteristic time scales for core dynamics that might be associated with excursion and reversal rate, time taken for reversals, or any signs of control by Earth's orbital parameters. The spectrum is characteristically red for the time interval 0-160 Ma, suggesting non-stationarity associated with average reversal rate changes, probably reflecting the impact of superchrons and a continually evolving core. Distinct regimes of power law decay with frequency may reflect different physical processes contributing to the secular variation. Evidence for non-stationarity at shorter time-scales is also present in dipole moment variations over 0-2 Ma with average growth rate faster than the decay process. Rates of change of dipole moment and rapid local field variations found in the paleomagnetic record are evaluated in the context of the 400 year historical record and the spectrum of geomagnetic variations for 0-160 Ma. (C) 2011 Elsevier B.V. All rights reserved.

Lawrence, KP, Tauxe L, Staudigel H, Constable CG, Koppers A, McIntosh W, Johnson CL.  2009.  Paleomagnetic field properties at high southern latitude. Geochemistry Geophysics Geosystems. 10   10.1029/2008gc002072   AbstractWebsite

Statistical analyses of paleomagnetic data from lava flows are used to study geomagnetic field behavior on million year timescales. Previous paleomagnetic studies have lacked high-latitude measurements necessary to investigate the persistence of geomagnetic anomalies observed in the recent and historical field and replicated in some numerical geodynamo simulations. These simulations suggest that reduced convective flow inside the tangent cylinder may affect the magnetic field at high latitude, whereas lower-latitude observations are expressions of columnar/helical flow outside the tangent cylinder. This paper presents new paleointensity and paleodirectional data from 100 volcanic sites in the Erebus Volcanic Province (EVP), Antarctica, and 21 new age determinations by the (40)Ar/(39)Ar incremental heating method. The new EVP data are combined with previously published paleomagnetic and geochronological results, providing 133 sites, 91 having radioisotopic dates. Modified Thellier-Thellier paleointensity estimates are reported for 47 sites (37 have dates). Ages for the combined data set span 0.03 to 13.42 Ma. The 125 high-quality EVP directional data selected from the merged data set have a non-Fisherian distribution and a mean direction with an inclination anomaly of similar to 3 degrees, but 95% confidence limits include the prediction from a geocentric axial dipole. Virtual geomagnetic pole (VGP) dispersions for Brunhes, Matuyama, and the combined 0-5 Ma data set are consistently high compared with values from middle-to low-latitude regions regardless of the criterion used to determine transitional fields. With VGP latitude cut off at 45 degrees, the dispersion (23.9 +/-2.1 degrees) for the combined 0-5 Ma EVP data set is consistent with earlier high-latitude data and paleosecular variation (PSV) in Model G but not with some more recent statistical PSV models. Mean EVP paleointensity of 31.5 +/-2.4 mu T, derived from 41 high-quality sites, is about half the current value at McMurdo (similar to 63 mu T). The result is essentially independent of data selection criteria. High VGP dispersion and low-intensity values support the global observation of anticorrelation between directional variability and field strength. Simulations of time-varying dipole strength show that uneven temporal sampling may bias the mean EVP intensity estimate, but the possibility of persistently anomalous field behavior at high latitude cannot be excluded.

Ziegler, LB, Constable CG, Johnson CL.  2008.  Testing the robustness and limitations of 0-1 Ma absolute paleointensity data. Physics of the Earth and Planetary Interiors. 170:34-45.   10.1016/j.pepi.2008.07.027   AbstractWebsite

Absolute paleomagnetic field intensity data derived from thermally magnetized lavas and archeological objects provide information about past geomagnetic field behavior, but the average field strength, its variability, and the expected statistical distribution of these observations remain uncertain despite growing data sets. We investigate these issues for the 0-1 Ma field using data compiled in Perrin and Schnepp [Perrin, M., Schnepp, E., 2004. IAGA paleointensity database: distribution and quality of the data set. Phys. Earth Planet. Int. 147, 255-267], 1124 samples of heterogeneous quality and with restricted temporal and spatial coverage. We accommodate variable spatial sampling by using virtual axial dipole moments (VADM) in our analyses. Uneven temporal sampling results in biased estimates for the mean field and its statistical distribution. We correct for these effects using a bootstrap technique, and find an average VADM of 7.26 +/- 0.14 x 10(22) A m(2). The associated statistical distribution appears bimodal with a subsidiary peak at approximately 5 x 10(22) A m(2). We evaluate a range of potential sources for this behavior. We find no visible evidence for contamination by poor quality data when considering author-supplied uncertainties in the 0-1 Ma data set. The influence of material type is assessed using independent data compilations to compare Holocene data from lava flows, submarine basaltic glass (SBG), and archeological objects. The comparison to SBG is inconclusive because of dating issues, but paleointensity estimates from lavas are on average about 10% higher than for archeological materials and show greater dispersion. Only limited tests of geographic sampling bias are possible. We compare the large number of 0-0.55 Ma Hawaiian data to the global data set with no definitive results. The possibility of over-representation of typically low intensity excursional data is discounted because exclusion of transitional data still leaves a bimodal distribution. No direct test has allowed us to rule out the idea that the observed pdf results from a mixture of two distinct distributions corresponding to two identifiable intensity states for the magnetic field. We investigate an alternative possibility that we were simply unable to recover a hypothetically smoother underlying distribution with a time span of only 1 Myr and the resolution of the current data set. Simulations from a stochastic model based on the geomagnetic field spectrum demonstrate that long period intensity variations can have a strong impact on the observed distributions and could plausibly explain the apparent bimodality. Our 0-1 Ma distribution of VADMs is consistent with that obtained for average relative paleointensity records derived from sediments. (C) 2008 Elsevier B.V. All rights reserved.

Korte, M, Constable CG.  2008.  Spatial and temporal resolution of millennial scale geomagnetic field models. Advances in Space Research. 41:57-69.   10.1016/j.asr.2007.03.094   AbstractWebsite

We assess the resolution and reliability of CALS7xK, a recently developed family of global geomagnetic field models. CALS7xK are derived from archaeo- and palaeomagnetic data and provide a convenient temporally varying spherical harmonic description of field behaviour back to 5000 BC. They can be used for a wide range of studies from gaining a better understanding of the geodynamo in the Earth's core to enabling the efficient determination of the influence of the geomagnetic field on cosmogenic nuclide productions rates. The models are similar in form to those derived from modern satellite observations, observatory and historical data, and used for the International Geomagnetic Reference Field, but their spatial and temporal resolution are limited by data quality and distribution. We find that spatial power is fully resolved only up to spherical harmonic degree 4 and temporal resolution is of the order of 100 years. Significant end effects associated with the temporal development in natural B-splines affect some features of the models in both the earliest and most recent century. Uncertainties in model predictions of declination, inclination and field intensity in general are smaller than 2 degrees and 1.5 mu T respectively, but can be as large as 8 degrees and 5 mu T for certain regions and times. The resolution studies are complemented by a detailed presentation of dipole moment and dipole tilt as predicted by the model CALS7K.2. These largest scale features are resolved more reliably than complex details of the field structure and are useful, for example, in studies of geomagnetic cutoff rigidities of cosmogenic isotopes. (C) 2007 COSPAR. Published by Elsevier Ltd. All rights reserved.

Korte, M, Constable CG.  2006.  On the use of calibrated relative paleointensity records to improve millennial-scale geomagnetic field models. Geochemistry Geophysics Geosystems. 7   10.1029/2006gc001368   AbstractWebsite

[ 1] Current millennial-scale time-varying global geomagnetic field models suffer from a lack of intensity data compared to directional data, because only thermoremanently magnetized material can provide absolute information about the past field strength. The number of archeomagnetic artifacts that can provide such data diminishes rapidly prior to 3000 B. C. Sediment cores provide time series of declination and inclination and of variations of magnetization: the latter can reflect relative geomagnetic field variations if suitably normalized. We propose a calibration technique based on predictions from global models and use the CALS7K. 2 model to calibrate relative paleointensity records from 22 globally distributed locations and assess whether they reflect actual field variations. All except a few contain useful information for 0 to 7 ka and could be used to improve the existing models. Using synthetic data from a numerical dynamo simulation, we show that with the existing directional data the distribution of intensity data has an important influence on model quality. Intensity data from a broad range of latitudes seem particularly important. This study opens the possibility of extending global time-varying geomagnetic field models further back in time than the current 7 kyr interval.

Korte, M, Constable CG.  2006.  Centennial to millennial geomagnetic secular variation. Geophysical Journal International. 167:43-52.   10.1111/j.1365-246X.2006.03088.x   AbstractWebsite

A time-varying spherical harmonic model of the palaeomagnetic field for 0-7 ka is used to investigate large-scale global geomagnetic secular variation on centennial to millennial scales. We study dipole moment evolution over the past 7 kyr, and estimate its rate of change using the Gauss coefficients of degree 1 (dipole coefficients) from the CALS7K.2 field model and by two alternative methods that confirm the robustness of the predicted variations. All methods show substantial dipole moment variation on timescales ranging from centennial to millennial. The dipole moment from CALS7K.2 has the best resolution and is able to resolve the general decrease in dipole moment seen in historical observations since about 1830. The currently observed rate of dipole decay is underestimated by CALS7K.2, but is still not extraordinarily strong in comparison to the rates of change shown by the model over the whole 7 kyr interval. Truly continuous phases of dipole decrease or increase are decadal to centennial in length rather than longer-term features. The general large-scale secular variation shows substantial changes in power in higher spherical harmonic degrees on similar timescales to the dipole. Comparisons are made between statistical variations calculated directly from CALS7K.2 and longer-term palaeosecular variation models: CALS7K.2 has lower overall variance in the dipole and quadrupole terms, but exhibits an imbalance between dispersion in g(2)(1) and h(2)(1), suggestive of long-term non-zonal structure in the secular variations.

Korte, M, Constable C.  2003.  Continuous global geomagnetic field models for the past 3000 years. Physics of the Earth and Planetary Interiors. 140:73-89.   10.1016/j.pepi.2003.07.013   AbstractWebsite

Several global geomagnetic field models exist for recent decades, but due to limited data availability models for several centuries to millennia are rare. We present a continuous spherical harmonic model for almost 3 millennia from 1000 B.C. to 1800 A.D., based on a dataset of directional archaeo- and paleomagnetic data and axial dipole constraints. The model, named Continuous Archaeomagnetic and Lake Sediment Geomagnetic Model for the last 3k years (CALS3K.1), can be used to predict both the field and secular variation. Comparisons and tests with synthetic data lead to the conclusion that CALS3K.1 gives a good general, large-scale representation of the geomagnetic field, but lacks small-scale structure due to the limited resolution of the sparse dataset. In future applications the model can be used for comparisons with additional, new data for that time span. For better resolved regions, the agreement of data with CALS3K.1 will provide an idea about the general compatibility of the data with the field and secular variation in that region of the world. For poorly covered regions and time intervals we hope to iteratively improve the model by comparisons with and inclusion of new data. Animations and additional snapshot plots of model predictions as well as the model coefficients and a FORTRAN code to evaluate them for any time can be accessed under The whole package is also stored in the Earthref digital archive at (C) 2003 Elsevier B.V. All rights reserved.

Everett, ME, Constable S, Constable CG.  2003.  Effects of near-surface conductance on global satellite induction responses. Geophysical Journal International. 153:277-286.   10.1046/j.1365-246X.2003.01906.x   AbstractWebsite

A 3-D finite-element simulation of global electromagnetic induction is used to evaluate satellite responses in geomagnetic dipole coordinates for harmonic ring-current excitation of a three-layer mantle overlain by a realistic near-surface conductance distribution. Induced currents are modelled for lithospheric and asthenospheric upper-mantle conductivities in the range sigma= 10(-4) -0.1 S m(-1) . The magnetic scalar intensity B is calculated at a typical satellite altitude of 300 km. At short periods, T = 2 and 12 h, the induction signal owing to the near-surface conductance is large when a resistive upper mantle is present, but drops off with increasing mantle conductivity. At longer periods, T = 2 d, the near-surface induction signal is generally much smaller and nearly independent of upper-mantle conductivity. The near-surface induction signal is very sensitive to the electrical conductivity of the lithospheric mantle, but only moderately sensitive to that of the asthenospheric mantle. Induced currents are confined to the heterogeneous surface shell at periods of less than 2 h, and flow predominantly in the mantle at periods of longer than 2 d. In the intervening period range, induced currents are partitioned between the near-surface and the upper mantle. These results indicate the importance of carrying out a full 3-D analysis in the interpretation of satellite induction observations in the period range from hours to days.

Lowe, DAJ, Parker RL, Purucker ME, Constable CG.  2001.  Estimating the crustal power spectrum from vector Magsat data. Journal of Geophysical Research-Solid Earth. 106:8589-8598.   10.1029/2000jb900437   AbstractWebsite

The Earth's magnetic field can be subdivided into core and crustal components and we seek to characterize the crustal part through its spatial power spectrum, R-1. We process vector Magsat data to isolate the crustal field and then invert power spectral densities of flight-local components along-track for R-1 following O'Brien et al. [1999]. Our model, designated LPPC, is accurate up to approximately spherical harmonic degree 45 (lambda = 900 km): this is the resolution limit of our data and suggests that global crustal anomaly maps constructed from vector Magsat data should not contain features with wavelengths less than 900 km. We find continental power spectra to be greater than oceanic ones and attribute this to the relative thicknesses of continental and oceanic crust.

RygaardHjalsted, C, Constable CG, Parker RL.  1997.  The influence of correlated crustal signals in modelling the main geomagnetic field. Geophysical Journal International. 130:717-726.   10.1111/j.1365-246X.1997.tb01866.x   AbstractWebsite

Algorithms used in geomagnetic main-field modelling have for the most part treated the noise in the field measurements as if it were white. A major component of the noise consists of the field due to magnetization in the crust and it has been realized for some time that such signals are highly correlated at satellite altitude. Hence approximation by white noise, while of undoubted utility, is of unknown validity. Langel, Estes & Sabaka (1989) were the first to evaluate the influence of correlations in the crustal magnetic field on main-field models. In this paper we study two plausible statistical models for the crustal magnetization described by Jackson (1994), in which the magnetization is a realization of a stationary, isotropic, random process. At a typical satellite altitude the associated fields exhibit significant correlation over ranges as great as 15 degrees or more, which introduces off-diagonal elements into the covariance matrix, elements that have usually been neglected in modelling procedures. Dealing with a full covariance matrix for a large data set would present a formidable computational challenge, brit fortunately most of the entries in the covariance matrix are so small that they can be replaced by zeros. The resultant matrix comprises only about 3 per cent non-zero entries and thus we can take advantage of efficient sparse matrix techniques to solve the numerical system. We construct several main-field models based on vertical-component data from a selected 5 degrees by 5 degrees data set derived from the Magsat mission. Models with and without off-diagonal terms are compared. For one of the two Jackson crustal models, k(3), we find significant changes in the main-field coefficients, with maximum discrepancies near degree 11 of about 27 per cent. The second crustal spectrum gives rise to much smaller effects for the data set used here, because the correlation lengths are typically shorter than the data spacing. k(4) also significantly underpredicts the observed magnetic spectrum around degree 15. We conclude that there is no difficulty in computing main-field models that include off-diagonal terms in the covariance matrix when sparse matrix techniques are employed; we find that there may be important effects in the computed models, particularly if we wish to make full use of dense data sets. Until a definitive crustal field spectrum has been determined, the precise size of the effect remains uncertain. Obtaining such a statistical model should be a high priority in preparation for the analysis of future low-noise satellite data.

Obrien, MS, Constable CG, Parker RL.  1997.  Frozen-flux modelling for epochs 1915 and 1980. Geophysical Journal International. 128:434-450.   10.1111/j.1365-246X.1997.tb01566.x   AbstractWebsite

The frozen-flux hypothesis for the Earth's liquid core assumes that convective terms dominate diffusive terms in the induction equation governing the behaviour of the magnetic field at the surface of the core. While highly plausible on the basis of estimates of physical parameters, the hypothesis has been questioned in recent work by Bloxham, Gubbins & Jackson (1989) who find it to be inconsistent with their field models for most of the century. To study this question we improve the method of Constable, Parker & Stark (1993), which tests the consistency of magnetic observations with the hypothesis by constructing simple, flux-conserving core-field models fitting the data at pairs of epochs. We introduce a new approach that fixes the patch configurations at each of the two epochs before inversion, so that each configuration is consistent with its respective data set but possesses the same patch topology. We expand upon the inversion algorithm, using quadratic programming to maintain the proper flux sign within patches; the modelling calculations are also extended to include data types that depend non-linearly on the model. Every test of a hypothesis depends on the characterization of the observational uncertainties; we undertake a thorough review of this question. For main-field models, the primary source of uncertainty comes from the crustal field. We base our analysis on one of Jackson's (1994) statistical models of the crustal magnetization, adjusted to bring it into better conformity with our data set. The noise model permits us to take into account the correlations between the measurements and requires that a different weighting be given to horizontal and vertical components. It also indicates that the observations should be fit more closely than has been the practice heretofore. We apply the revised method to Magsat data from 1980 and survey and observatory data from 1915.5, two data sets believed to be particularly difficult to reconcile with the frozen-flux hypothesis. We compute a pair of simple, flux-conserving models that fit the averaged data from each epoch. We therefore conclude that present knowledge of the geomagnetic fields of 1980 and 1915.5 is consistent with the frozen-flux hypothesis.