Summary
The main target formation of Field A is a lowstand wedgeshaped delta sandstone developing on the shelf-break belt, which has thin layers and fast lateral variation. On the top of sandstone, there is a set of ultra thick mudstone with high speed. So it is difficult to describe sandbody accurately. This work gives a high-precision attributes analysis technology to resolve the problem. We extract multiple high-precision attributes based on analysis of reservoir characteristic, and then make fusing display using K-L transform results, to describe development zones of reservoir. At last, we use the neural network method to get the distribution of reservoir, and its statistical thickness. The results are satisfactory compared with well data.
Introduction
The rule of global exploration shows that the shelf-break belt and lowstand tract which under its control are most conductive to the formation of unstructural reservoirs.
Field A is located in the east of the South China Sea. The target formation (S1) is a lowstand wedge-shaped delta which developing many sets of foreset sandbody with wider distribution and fast lateral variation. On the top of sandstone, there is a set of ultra thick mudstone with high speed, which leads to a complex wave reflection formed by the bottom interface of mudstone and the top interface of sandstone. It is so difficult to describe sandbody accurately, let along low-resolution data.
In this paper, we use high-precision attributes analysis technology in sandbody description of S1. First, we extract two attributes with high accuracy, which are edge detect attribute based on high dimensional continuous wavelet transform (HCWT) and high-precision trace integral using multistage wavelet(MWTI), and other attributes which related to reservoir distribution on the basis of analyzing reservoir characteristics, such as amplitude, frequency, structure, and so on. Then we use K-L transform for dimensionality reduction after sensitivity analysis, to obtain the preferred subsets of attributes. At this time, we can describe the development zones of reservoir qualitatively by using RGB fusion technology. Furthermore, we obtain the 3D pseudo-GR data by using the method of PNN neural network and lithologic body thickness by using the statistical method.
