How to Cure the Shortsightedness of the Seismologits and the Indescisiveness of the Gravimetricians?

Geophysical, geological and geodesic data have been inverted for decades now to provide the most reliable image of the Earth at all scales. This information is essential to any tectonic interpretation or global model. Still, due to the type of data, their limited number, their uneven distribution and the measurement errors, geophysical inversions are very often non-unique and/or yield to geologically unrealistic models. Many schemes have been undertaken to address these problems. Since about a decade now geophysicists have used "cooperative inversions" as first mentioned in Lines et al. (1988) to overcome problems mentioned above. The basic idea of this method is to gather complementary data within the same inversion scheme in order to reduce the model set. The first approach is called sequential, where the data sets are considered in separate inversions. For example, one solves the inverse gravity problem for density with pre-assigned geometry deduced from seismic studies. The second approach is called joint inversion, and assumes a known relationship between the different parameters to invert. In this case, the data are inverted simultaneously (e.g. inversion of magnetic and gravity data under assumption that the anomalies are caused by the same sources). We discuss here the advantages of these approaches by considering the example of a joint inversion of gravity and teleseismic delay times in an iterative scheme. Variation of density and P-wave velocity are linked via an empirical law (e.g. Birch, 1961). This technique was first explored in the 80's and recently showed interesting results in many regions of geodynamic interest (Baikal, Moluca Sea, Greece, East African Rift). An alternative way of considering gravity, seismology and any other data is through geodynamical models. In this approach, a geodynamical model, which describes a structure or tectonic process under study, contains a set of parameters. Synthetic gravity and seismological data are then functions of the same set of parameters. They can be calculated in a sequential way and compared with real observations. We will illustrate this method with the example of the Sumatra earthquake. Present day or future high accuracy data from dedicated space gravity missions such as GRACE and GOCE open the possibility of knowing much more about the structure and dynamics of the Earth, with greater details, particularly through joint analysis with seismological data. We discuss here the ins and outs of such cooperative inversions.