Robust mean traveltime curves: a powerful tool for analysis in 2D transmission tomographic surveys

Acoustic tomography methods belong to the class of non destructive inspection techniques and are widely used in engineering applications. One of the main issues for these methods is the direct arrivals, which can be noisy or affected by scattering or other propagation effects. Mean traveltime curves (mean and standard deviation) have been introduced by Fernández Martínez et al (2006), and describe the variation of the main statistical parameters of the traveltime distribution for the different gathering subsets as a function of the gather index. These curves constitute a simple methodology (Fernández Martínez et al., 2008) to analyze before inversion the traveltime quality in 2D acoustic transmission tomography experiments conducted in relatively homogeneous blocks. This analysis can help to discriminate the contribution of picking errors from that of geological heterogeneities. In this contribution we present the so-called robust mean traveltime curves, which refer to the variation of the different p-percentiles and related measures of dispersion (inter-quartile range and minimum absolute deviation) for the above mentioned gathering subsets. Together with the mean curve, the median and upper-quartile traveltime curves are very resistant to the presence of outliers, and thus, are preferred to infer a background velocity model, which is valid at experiment scale and honors the experimental traveltime distribution. This methodology is valid for 2D zonal isotropic or elliptical anisotropic media, explored via any arbitrary acquisition geometry. We show the application to a well-known granitic medium (Febex Project, Nagra, Switzerland). Fernández Martínez et al. 2006. Mathematical Geology:38-3.343-374 Fernández Martínez et al.2008. Computer and Geosciences:34-3,213-225, 2008.