The Reliability Research Of Seismic Wave Attributes

In this paper, The theory and application of how to abstract and classify the seismic wave attributes are discussed. The reliability of seismic attributes has been being the main objective on which geophysicists make their efforts for long time. The category of seismic attributes and the method of how to abstract and classify the seismic attributes is different. How to abstract the accurate and reliable attributes is become the key point.In this paper, the necessity and significance for carrying this thesis was discussed firstly. The lateral and longitude resolution of seismic wave and how to abstract the attributes are discussed in detail. Forward model is an effective method that we discuss the seismic attributes and geophysical question. Application the forward model we can get the profile of medium that combine in some condition, In order to conduct the attributes optimize, the paper offer some basic principle, and discuss the reliability of attributes from the character of incident wave and from the character of the inter-bed and from the reliable zoon of attributes.Application the theorem of cluster analysis, the classification of attributes is discussed. Some attributes have the same or likeness feature, so them reflect the same or likeness geophysical phenomena. In field works, when we using the attributes that haven't classify, it will cause some information be used many times and some information will not be used. Cluster analysis using the conception of distance parameter, made the attributes that have the high correlativity become the same sort. So the classification of attributes become complete and correct. Now the application of attributes can reflect the phenomena of geophysical completely and concisely.The characteristics of seismic signal are influenced by the rock and the physical character of liquid directly. It is very useful to excavate the information about rock and deposit from the seismic data. It's a nonlinear relationship among the seismic attributes, and the influence of environment is obviously. BP algorithm can establish the high nonlinear mapping among the object and seismic attributes. GA algorithm can select the survival of the fittest. The BP algorithm can quickly converge, and the GA can global search the fittest result in all data. Apply the GA+BP hybrid algorithm in inter-bed prediction, and using the attributes combination that come form cluster analysis we can achieve a satisfactory result.