Locating deep reservoirs with economic potential in the Texas Gulf Coast using AVO

Abstract

Sandstone reservoirs deeper then 3200 m (10,500 ft) are potentially rich sources of hydrocarbon in the Texas Gulf Coast. This study focused on deep potential reservoirs in the Edna Field, located in Jackson Co, TX. Due to lack of well control, Amplitude Variation with Offset (AVO) analysis was used on 2 sands located between 10,500 ft and 14,000 ft to investigate potential hydrocarbon charge and to validate if the bright spots are hydrocarbon associated. Before AVO analysis began, a prestack time migration (PSTM) processing flow was applied to the dataset to assure proper reflector alignment in the prestack gathers. This is done to optimize the AVO analysis. Schlumberger's processing package Seismos® was used for the processing. In Seismos®, the f-k migration module was used to produce two data volumes: a Multi-Value Grid (MVG) and a Single Value Grid (SVG). A MVG allows for multiple values to exist at the same x-y-z point in space. This allows for each offset plane to be analyzed. A SVG has one value at each x-y-z point in space. This represents a conventionally stacked seismic volume. After PSTM, the dataset was repicked to update sand geometries. As with the seismic volume before reprocessing, both sands contained bright spots that conformed to structure. With the PSTM dataset and updated interpretation, both sands were analyzed for the AVO anomalies. On each sand, standard intercept (A) and gradient (B) techniques were performed (gradient analysis, product plots, change in Poisson's ratio). In addition to this, the Geostack technique of Smith and Gidlow (1987) was done. The A&B techniques and Geostack technique for the 10,500 ft sand resulted in anomalies that agreed with all the bright spots in the conventionally processed data. The results for the 14,000 ft sand were more ambiguous. This uncertainty was due to lack of offset for depth and low fold at the edge of the survey where the sand was positioned. Intercept and gradient techniques did not support the location of the bright spots in the conventionally processed data.