Summary
Between 1997 and 2003 the Gemini Prospect, Gulf of Mexico, was used as a test ground for development of broadband marine magnetotellurics (MT). In 2003 controlled source electromagnetic (CSEM) data were acquired over Gemini using a prototype 200A transmitter on a profile that included earlier MT sites. In 2013 and 2014 these data were revisited as part of the evaluation of the recently developed Scripps 2D EM inversion/forward modeling code, MARE2DEM, developed largely by K. Key. A portion of the results of this evaluation are presented here with the objective that of examining aspects of the application of the inversion algorithm to real CSEM and MT data, rather than a detailed re-interpretation of the Gemini data set. Of order of 100 inversions were carried out as part of this study. The processed CSEM data were inverted for each of three frequencies individually and then for the three frequencies jointly, the MT data alone were inverted, and finally the multi-frequency CSEM and MT data were jointly inverted. Each of these data sets was subject to isotropic inversion, and then the joint CSEM/MT data were the subject of anisotropic inversion. Target misfits were varied and the results compared. Inversions were carried out with a half-space starting model as well as with more complex starting models. The results provide insight into the inversion outcome based on the effects of choice of data component to be used as input, the choice of starting model, target misfit selected, and the degree of anisotropy. The results imply that for CSEM and MT data collected at multiple frequencies, inverted models can change radically as different subsets of data are inverted individually or in combination, and with varying target misfit and/or anisotropy penalty.
