Analysis of Fluvial Crevasse Splay Geometries from Satellite Imagery and 3D Seismic Data
Masters of Science Degree 2008
University of Adelaide
Fluvial architecture is very important in the oil and gas industry because sandy fluvial deposits often form reservoirs for oil and gas. As detailed data and quantitative data sets of fluvial systems in petroleum companies are often lacking, it is necessary to generate fluvial analogue data sets in order to produce a statistical analysis. This study focuses on one element of fluvial systems, the crevasse splay.
This study uses satellite images of a selection of rivers from around the world and 3D seismic data in the Pattani basin, Gulf of Thailand. Interpretation of the statistical analyses and analogues are made based on the dataset.
This study develops a descriptive and quantitative data set of fluvial crevasse splay geometries. More than 100 splays are analyzed to describe the basic trends from cross plots of data. The relationships between these parameters are explored. It can be concluded from the plotted graphs and seismic time-slices that the size of unconfined splays that are found along the channel margin appear to be more laterally extensive than levee splays and the splays interpreted from seismic. The large crevasse splay is most likely to be found associated with the wider channel because more sediment is carried due to the increase of velocity and discharge. The unconfined splays appear to occur in association with the wide and low to moderate sinuosity channels, whereas levee splays are found along the relatively narrow-to-medium sized and low to moderate sinuosity river channels. Moreover, the smaller size of the levee splays could possibly be explained because the sediment which is carried by a main river channel can be deposited along the crevasse channel before it is deposited as a crevasse splay lobe.
The results from this study will assist business units worldwide in the determination of realistic crevasse splay geometries to be used in the 3D reservoir modeling.