69 3. Seismic tomography The picked travel times were used to compute the refractor velocity at the base of the epikarst. It was noted that the values change between 2,000 m/s and 3,200 m/s, with a clear azimuthal variation, being the strike of the fast axis around 175°N (Figure 3.5). Despite the azimuthal variation of velocity, for simplicity, an initial 1D velocity model, with linear velocity gradients, was considered adequate for the tomographic inversion. Then, the input data were: • 7,813 first-break times; • 1D model with 2 linear vertical velocity gradients; • 100 Hz central signal frequency. 3.1.4 Results and discussion The processing of seismic data led to a 3D velocity block. The inversion scheme was stopped after 10 iterations, when the misfit function reached 1.69 ms for the root-mean-squared (rms) error, i.e., the squared difference between measured and calculated travel times. An example describing the quality control of the inversion results is specified for shot number 83. Figure 3.6 shows three travel time sets: a) the picked times of field data; b) computed times for the initial model; and the times for the best model provided by the tomographic inversion. The evolution of travel time residues, during velocity model building, is illustrated in Figure 3.7. Note the significant reduction of travel time differences, showing a good tie between the picked and best model data. This provides assurance that the final model accurately reflects the field data. Figure 3.6 Diagrams of travel times for shot number 83. The picked times of field data (black triangles), travel times for initial model (asterisks) and travel times for a model provided by iteration 10 (crosses).
RkJQdWJsaXNoZXIy NjA3NzQ=