:
Seismic multi-attribute analysis technique has been widely used in reservoir prediction. Beach bar sandstone, alternating with mudstone, is difficult to distinguish using conventional seismic data because of its small individual bed thickness. In the process of seismic attribute analysis, two problems exist: 1) difficulties to ensure the geologic meaning of seismic attributes; 2) choosing time window properly. In the study of Es4 beach bar sandstone in Liang108 Area of China’s Dongying basin, similar background separation technique (SBST) is used to make the reflection from beach bar sandstones standout more from background, and layer slice technique (LST) and super trace analysis technique (STAT) are developed to improve the precision of seismic attribute analysis. With these problems addressed properly, seismic attributes are more correctly and efficiently used in the process of reservoir prediction.
:
Seismic attributes, extracted from seismic traces, can be used to not only predict reservoir by the application of pattern recognition technique, such as neural net and statistical analysis, but also directly detect hydrocarbons under some conditions (Zeng, 1997). The steps of a typical seismic attribute analysis are: first, properly choose seismic data which stands for reflections from target stratum and extract all kinds of attributes; second, make cross-plot analysis of these attributes, use well log information to group these attributes; finally assign geological meaning to different attribute group. During the last 10 years, many methods and programs have been developed for seismic attribute analysis. Most work was put on the extraction of new seismic attributes and on the methodology of attributes classification (Gao, 2000). Now seismic data could be used to extract attributes related to amplitude, frequency, phase, energy, wave shape, and so on. Current software can extract more than 60 different kinds of attributes, and new kinds of attribute being developed and used continuously. Today the research of attribute classification based on reservoir characterization is very active. For example, there are classification methods based on fuzzy mode recognition, neural net, function approximation, geo-statistics and combinatorial method (combine previous methods). However, the validity of attribute extraction, especially when extracting thin bed information, is often ignored. In the seismic attribute analysis study of beach bar sandstones in Liang108 area, SBST, LST and, STAT are used to make the reflection from beach bar sandstones more standing out, and to improve the reliability of seismic attribute analysis.
The main reservoir of Liang108 Area is beach bar sandstone in sand group I of Es4. Thickness of sand group I varies from 30m to 60m. Single sandstone bed thickness changes from 1m to 3m and lateral change is acute. The sandstone is interbedded with mudstone and limestone (Cai, 2005). In the process of using seismic information to predict beach bar sandstone distribution, the main objectives become how to choose attributes that could make the reflection from the target stratum more standout and how to choose proper analysis time window.
