E. Herrero-Bervera $^{a}$ and J-P. Valet $^{b}$
$^{a}$ SOEST-Hawaii Institute of Geophysics and Planetology, University of Hawaii, Honolulu, Hawaii, USA. $^{b}$ Institut de Physique du Globe de Paris, Paris, France.
Most recent studies of the geodynamo emphasize the importance of documenting the time evolution of the geomagnetic field over million of years in order to impose constraints of the physical processes that are responsible for field generation in the liquid outer core. One of the key features of the goemagnetic field is its ability to reverse polarity with several hundred reversals having occurred over the last 160 Ma. The mechanism of reversal processes is as yet proorly understood. Reversals probably take a finite time but are very short on the geological time scale. Typical time scale for transitions range from 1000 to more than 10,000 years and there are indications that different reversals could be well characterized by very different durations. During the transitional process the field appears to be considerably reduced, probably leaving the rapidly varying non-diplole components emerging at the Earth's surface and thus making this period difficult to study. Knowledge of the mechanisms that are responsible for field generation require detailed records of the total field vector, and thus records that incorporate directional and intensity changes. Analysis of the total vector will provide us with significant information by constraining the non-dipole and the dipolar components with their charcteristics time constraints. Proper documentation is also required to test the field models derived from numerical simulations of the dynamos with the actual characteristics of the geodynamo. So far most reversal studies have been dealing with sediments. However, criticisms and skepticism have been raised regarding the ability of sediments to accurately record the field changes during periods of reduced geomagnetic intensity. Another drawback inherent to sedimentary sequences is that they restore a time-averaged signal, which is not too critical during periods of full polarity but becomes a major drawback when studying rapid field changes associated with the reversals. Despite discontinuous temporal recording, lava sequences thus appear to be appropriate to document the field variations that are inherent to the emergence of non-dipole components during transitional and excursional periods. Indeed lava flows can potentially provide estimates of absolute field intensity and thus a proper documentation of the total field vector. Clearly, both sets of data are complementary and both are thus necessary. Ideally what is needed is a volcanic sequence that would cover a long time period recording successive reversals and thus bearing similarities with sediments. So far, no such record has been obtained, and this limits our knowledge of the field variations accompanying reversals and the surrounding polarity intervals. The most detailed volcanic record from the Steens mountain includes only one reversal and does not document the polarity intervals preceding and following the reversal. It remains that this record is still suprisingly the best documented reversal record that has been ever reported in the literature. The Wai'anae volcano lavas (3.8-2.4 Ma) located on the west part of the island of O'ahu, Hawaii has provided a unique opportunity to document the geomagnetic field behavior over a relatively long period of time. This has been an ideal location to investigate the Earth's magnetic field because extensive and superbly well exposed lava flows make up sequences that includes at least three back to back reversals of the geomagnetic field (Gilbert-Gauss 3.57 Ma through the Upper Mammoth 3.22 Ma). Five thick sections appropriate for paleomagnetic sampling have been successfully studied. The geomagnetic results include the pre-and post-transitional directions with determinations of absolute paleointensity from the three successive reversal volcanic sequences. The dominant pattern of the directional changes is the presence of large inclination paleosecular changes with increasing amplitude and steep values during transitional periods. Our results indicate that there is no evidence for preferred longitudinal sectors nor for cluster of directions that would be related to periods of stanstill during reversals . In contrast the presence of parallel records has demonstrated that clusters of directions are not reproducible between parallel sections and thus related to local eruption rates. Some recurring features could reflect characteristics inferred from studies of the secular variation in the Pacific. Preliminary absolute paleointensity experiments of 240 lava flows have been conducted on one sample per lava flow using a modified version of the Thellier-Coe technique. We obtained a 23 percent success rate for the paleointensity determinations and indicate that there is a range between 3.5 and 88uT. The paleofield was reduced to at least one tenth of the present day value during transitions characterized by a mean paleointensity of 15uT whereas the field recovery depicts a mean paleointensity of 40 uT (normal polarity intervals) with severe paleointensity highs during the recovery periods.