Abstract
The Earth's magnetic field shows a large range of temporal variations from the year to several million years. I studied during my PhD thesis several aspects of these fluctuations, from recent secular variations to reversals. The first part of my manuscript deals with archeomagnetism, a discipline that allows to track the temporal variations of the Earth's magnetic field through millennia, mainly from archeological materials. I studied two sets of samples , one from Greece (Neolithic period 6800-3200 B.C.) and the other from Mexico (Palenque, Maya Classic period 320-840 A.D.), to acquire new archeointensity data in order to better constrain the secular variation of the geomagnetic field. By comparing my data with those available in the literature and with the various global and regional models, I showed that the secular variations during the Neolithic in Greece and during the first millennium in Central America are poorly defined. My data suggest that local components, not described by global models, may exist in these regions. They reinforce the importance of developing specific regional models. As a consequence, the acquisition of new high quality data is of main importance. The second part presents the description of a geomagnetic field reversal. This work was based on two points : first by studying transitional directions to provide new constraints on the possible preferred longitudinal paths of virtual geomagnetic poles (VGPs) and second by checking transitional paleointensities obtained on a volcanic sequence in Akaroa volcano (New Zealand). Indeed the transitional field intensity is significantly higher than the field intensity before and after the reversal. We re-sampled this sequence, and the directional results show a complex sequence of N-T-R-T-N-T-R polarity. The transitional VGPs obtained are clustered in two longitudinal bands through Australia and America. This observation seems to reinforce the assumption of a core-mantle interaction over several million years. Following a rock magnetic study, I selected samples that could provide a value for the Thellier and multispecimen paleointensity methods. The obtained paleointensity are relatively low (about 20 μT) during the polarity change and strong at the end of the sequence. Based both on the field strength values and on the radiochronological ages, showing that the sequence was erupted in a very short time, I suggest that only the C4AR.1n-C4Ar.1r reversal was recorded in this sequence. In this assumption, the reversal shows a complex path comparable to other reversals recorded in the northern hemisphere (for example the Steens Mountain ), including a rebound before stabilizing. Finally the last part is devoted to the development of a new methodology and a new apparatus to determine absolute paleointensity. Following the low success rate of paleointensity experiments from the previous study, I decided to test the multispecimen protocol, which can be applied to samples yielding a predominant multidomaine behavior. The main technical drawback of this method lies in the application of the laboratory field along the natural remanent magnetization, a difficult task to perform accurately in standard paleointensity ovens. Thus, we decided to adapt sample holders from our standard oven in order to allow the sample orientation in space and to develop an ultra-fast heating oven prototype particularly well-suited for this method, allowing to apply the laboratory field in the 3 dimensions. I checked the different multispecimen protocols on historical lavas from Reunion and Etna volcano, yielding very different magnetic mineralogies. For all flows, I obtained paleointensities very close from the expected values, regardless from the magnetic mineralogy, revealing the feasibility of our apparatus and the promising interest of the method. The application of various corrections on the statistical estimation of the intensity value or on the cooling rate had no significant impact on my results.