地震属性处理演示文稿.ppt

Lightscape Lightscape Lightscape Lightscape Lightscape * Coherency : linked to the local continuity of seismic horizons * First, I will show you explanations in2D for a seismic vertical section. The dip is defined as the angle associated to the translation time enabling the best local correlation between seismic traces. We consider seismic locally as rectilinear segments, that implies isovalues of seismic data are lines. The gradient g is defined as the vector with coordinates equal to vertical and horizontal local variations of seismic data. On an isovalue curve, the gradient is orthogonal to the curve itself. Then the gradient direction gives, the seimic dip to within 90 degrees. * The gradient value at a given point is not sufficient for determining the local dip. It is necessary to work with a neighbourhood around a point. We then use gradients at all points of the neighbourhood N,and we are looking for the unitary dip vector minimising the dot product between these gradients and this unitary dip vector, i.e. is orthogonal in average to all these gradients. * The coherence is estimated using both eigenvalues. More close are eigenvalues, C is closer to 0, and seismic data are badly correlated within the dip direction. On the contrary, if lM is more greater than lm, C is close to 1, and seismic data are better correlated within the dip direction. C can be seen as an indicator of confidency for the dip computation. * This is equivalent to perform a non-normalised principal component analysis of the local gradient vector field, or to perform a linear regression on gradient components. * In order to take into account boundary effects with uniform summations on a rectangular neighbourhood, the expressions have been modified. For this, we replacethe sums by the application of a filter F(x,y) which performs a weighted average on the whole image. Fo example, the dip is : with x the 2D convolution . The filter chosen is an exponential filter with a recursive impleme