Direction and intensity of Earth's magnetic field at the Permo-Triassic boundary

The Earth's magnetic field is generated by the motion of liquid iron-rich material in the outer core. One of the most drastic manifestations of the dynamics in the outer core are polarity reversals of the magnetic field. The processes controlling geomagnetic reversals, however, are still poorly understood. The mathematical formulation of the dynamics of the liquid outer core show such a degree of complexity that a universal numerical model still remains elusive. Given that the last reversal occurred about 780,000 years ago, direct observations of a reversal have never been possible. Thus we are left with records of ancient reversals recorded in sequences of sedimentary and igneous rocks. Documenting any systematics in reversal processes will provide substantial information about the outer core and core mantle boundary conditions. However, despite the advances in deciphering the behaviour of the field during polarity transitions, reversal records yield controversial results and thus answers to several key questions are still enigmatic. Detailed studies of palaeodirectional and absolute palaeointensity patterns of geomagnetic reversals are scarce and are restricted to the Cenozoic so far. In order to verify or reject concepts developed on the basis of this dataset, reversal records which occurred in the more distant geological past of the Earth are needed. This work presents the results obtained from the Siberian Trap Basalts (Russia) which are coeval with the Permo-Triassic boundary (250 Ma). The sequence yields the by far oldest hitherto studied detailed record of a geomagnetic transition from reversed to normal polarity and provides new insights in transitional field behaviour. Three sections (Talnakh, Listvjanka and Abagalakh) comprising a total of 86 lava flows have been sampled in the Noril'sk region, located at the northwestern rim of the Siberian Trap Basalt province. They provide a complete coverage of the lava pile outcropping in the area. The samples have been subjected to palaeomagnetic direction analysis and to Thellier-type palaeointensity experiments. Extensive rockmagnetic investigations and microscopical studies have been carried out to asses the reliability of the palaeomagnetic information recorded by the lava flows. Magnetite and Ti-poor titanomagnetites were identified to be the carriers of the characteristic remanent magnetisation. The reversibility of the thermomagnetic curves and the observation of exsolution lamellae by ore microscopy give clear evidence for a primary high-temperature oxidation of the titanomagnetite. It can thus be inferred that the measured palaeodirectional and intensity information obtained from these flows was acquired shortly after extrusion of each flow. The demagnetisation of the natural remanence reveals only one direction of magnetisation for most samples. Thermal and alternating field demagnetisation methods are equally effective in isolating the characteristic remanent magnetisation. Occasional overprints have maximum unblocking temperatures of 350°C or remanence coercivities less than 20 mT. Reliable palaeointensity estimates were obtained for approx. 50% of the samples. The relatively high success rate can be attributed to the enhanced magnetic and thermal stability of high-temperature oxidised titanomagnetites. In the lower part of the sequence reversed polarity of the Earth's magnetic field is identified. The associated palaeointensities yield values around 10 µT. The subsequent flows recorded transitional configurations. A tight cluster of virtual geomagnetic poles (VGPs) in mid northerly latitudes, comprising the results of 15 flows, is observed during the transition. Within the cluster the record shows a pronounced and well defined increase in intensity from around 6 to 13 µT. A doubling of local field intensity infers that large scale dynamic processes in the outer core are responsible for this feature, making a strong case for a reasonable temporal stability (several hundreds to a few thousand years) of the VGP cluster. Moreover, the VGP clustering is identified in two parallel sections (Talnakh and Listvjanka). This observation makes it unlikely that this feature is an artifact of a localised burst in volcanic activity and supports the concept of stabilised phases of the geomagnetic field during reversals. The VGPs of the overlying flows move towards the position expected for normal polarity. After rotating of the VGPs into the Late Permian/Early Triassic geographic reference system it is evident that most of the transitional VGPs are strongly confined to a narrow longitudinal band which is perpendicular to near- or far-sided VGP paths. Such near- or far-sided paths would be indicative for the dominance of zonal, and thus axis-symmetric, non-dipole fields. The VGP path of this transition suggests the contribution of strong sectorial components of the Earth's magnetic field. Following the transition itself, normal polarity is reached for a brief time interval. Subsequently, the VGPs depart from this position to form another well defined directional cluster recorded by 14 successive flows. During this clustering, which is interpreted as an excursion of the Earth's magnetic field, no characteristic variation in palaeointensity is identified (mean value 14 µT). Such post-transitional excursions are frequently observed in younger reversal records and are explained by instabilities of the geodynamo after the reversal. However, VGPs associated with post-transitional excursions usually reach positions similar to those occupied by VGPs during the transition. In contrast to such "rebound" effects, the excursion-related VGPs of this record are still confined to the latitudinal band defined by the transition, but "overshoot" normal polarity. This geometrical constraint suggests that non-dipole components similar to those dominating the transitional VGP path are responsible for this observation. Remarkably, the geomagnetic polarity transition described here shares many similarities - such as directional clustering, longitudinal confinement of the VGP path, the existence of a post-transitional excursion and generally low palaeointensities - with previously published reversal records of mainly Tertiary age. It may, therefore, be inferred that the underlying reversal processes are similar to those observed for the Cenozoic. The results obtained of the superjacent 41 flows, which were extruded immediately after the reversal-related excursion, indicate that only at this stage of the record stable normal polarity is reached allowing to determine several characteristic parameters of the Early Triassic Earth's magnetic field. The mean palaeointensity for this part of the sequence is 19 µT, which corresponds to a virtual geomagnetic dipole moment (VDM) of 2.3 * 10^22 Am^2. These findings confirm that the Mesozoic dipole low extends at least down to the Permo-Triassic boundary. Calculation of the recorded secular variation yields values similar to those averaged over the last 5 Ma, a period with distinctly higher mean VDM (5.5 * 10^22 Am^2) compared to the data presented here. The hypothesis of enhanced secular variation during phases of a low mean VDM can, therefore, not be substantiated by this study. Secular variation and the strength of the dipole moment seem to be - at least in the Early Mesozoic - more complexly coupled than previously assumed. Magnetostratigraphic results of borehole samples obtained from basalts related to the Siberian Trap volcanism including the West Siberian basin yield in total 6 polarity intervals. Comparison to the global magnetostratigraphic scale indicates that the volcanic activity lasted no more than 3.2 Ma. However, the lava sequence in the Noril'sk area (more than 1700 m thick), representing the bulk of the erupted material, recorded only one polarity transition. This finding has been supported by data derived from boreholes in close vicinity to the surface sections which makes the presence of further undetected polarity transitions highly unlikely. It can be thus inferred that the emplacement of the sequence occurred much faster than the aforementioned 3.2 Ma. Radiometric ages suggest an upper limit for the duration of the emplacement of approximately 1 Ma. Based on the assumptions of similar rates of angular secular variation in the Early Triassic and in the Holocene and an average duration of the transition itself the time interval covered is estimated to be in the order of 15000 years. This value has to be regarded as a lowermost limit for the duration of the emplacement. Such a rapid development of the volcanic province in the Noril'sk area would imply an enormous eruption rate making a strong case for the Siberian Trap basalts as cause for the Permo-Triassic crisis