Interval velocity analysis and reflection tomography for resolving near surface velocity: A case study

Abstract

Summary A case study is presented which applies interval velocity analysis and reflection tomography to real 2D land seismic data in an area where the primary problem is optimizing the depth conversion of low relief time structures in the presence of subtle lateral velocity gradients. The technique utilizes commercially available velocity analysis and Pre Stack Depth Migration (PSDM) software. The example demonstrates that reflection tomography can resolve shallow velocity anomalies that can not be resolved by either stacking velocity analysis or interval velocity analysis. In the case presented here velocity problems were resolved by using as input for tomography the moveout errors of deeper horizons obtained from an initial pass of PSDM. Once the shallow velocity model was resolved, the interval velocity analysis of subsequent layers became stable and a more refined velocity model was constructed. After a number of iterations of PSDM and reflection tomography, a final 2D velocity model was derived. The chosen seismic line contains three wells which were used to judge the quality of the final depth migration. The structural framework of the Surat Basin is controlled by the configuration of the underlying Bowen Basin which underwent a period of significant compression, uplift and erosion during the Late Triassic. Subsequent periods of reactivation during the Late Jurassic, Late Cretaceous and, ultimately, the Tertiary have accentuated pre-existing structures. Traps are typically low relief anticlines, often associated with Triassic faulting, with structural dips of 0 to 10 degrees. Seismic imaging of the major reflectors is generally good to excellent, however subtle lateral velocity gradients introduce time-depth discrepancies that have led to the drilling of time highs which are not valid depth closures. Presumably also, depth closures exist in these basins that have no expression in time. During field development velocity gradients commonly cause problems for depth prediction especially when addressing attic reserves. The primary cause of the time-depth discrepancy appears to be two-fold: • A thick weathered layer introduces shallow lateral velocity gradients that are at best imperfectly estimated by datum static calculations (usually refraction statics tied to uphole surveys). • Subtle velocity gradients within deeper layers generally go undetected because they are masked by the shallow anomalies which perturb the deeper stacking velocities. Caxton 1(Well C), located on the southern extension of the greater Kincora structure, was drilled in 1995. The objective of the well was to test a fluvial channel directly below marker H4 on the seismic section. The well was located on a time high that was mapped on an old line. Depth conversion of the target horizon (H4) was effected by using an average velocity profile derived from nearby wells. In the absence of any anomalous average velocity trends, the resultant depth high was coincident with the time high and a fault-independent anticlinal closure of approximately 19 metres relief was mapped. Caxton 1 intersected the target 22.7 metres low to prognosis, but addressed significant reserves of stratigraphically trapped gas outside conventional closure. The result at Caxton 1 pointed to the likely existence of statics problems and/or a significant velocity variations. Horizon Stacking Velocity Analyses (HSVAs) were computed for horizons H-1 to H-4. These analyses exhibited oscillatory profiles which became more pronounced in successive layers. This is a common feature of stacking velocity profiles in the Surat Basin and the Eromanga Basin of central Australia. It is the reason why stacking velocities are not normally used for depth conversion; they vary much more widely than the well control demonstrates.