Resolution, Detection and Quantification

Abstract

The use of geostatistical inversion is often regarded as the best option for exploring for thin beds. This is based on the belief that significant enhancements in resolution can be achieved. However, there is little evidence for such improvements over standard inversion. Knowledge of the properties of the rocks when used in joint inversions for categorical and continuous data will improve resolution. However, the presence of noise can cause significant deterioration in resolution and detection of both thin and thick beds. Detection is controlled by signal to noise and on a trace by trace basis the limit of detection is when there is of the order of 2 and 3 standard deviations of the noise levels between the background and the signal. Detection can be improved by the recognition of connected bodies with thresholded properties. In all cases simple deterministic inversion improves detection and resolution of thin beds over the use of seismic amplitudes by a factor of about three. Joint inversion can improve the thin bed resolution further but not by a large factor.