达里亚工区速度体建立中几种网格化方法的对比分析

Gridding methods used for constructing velocity volume in Daliya: Comparative analysis

地矿部石油物探研究所, 南京210014

Research Institutc of Gcophysicak for Pctrolcum, MGMR, Nanjing 210014

本文着重对比分析了四种网格化方法, 即近点按距离加权平均法(N－P法)、按方位取点加权法(带方向的N－P法)、趋势面拟合法及趋势面加残差叠加法。通过试验可清楚看到, N—P法和按方位N—P法都属插值类, 这类方法不改变原始测量点值, 而且运算速度快, 但N－P法没有考虑方向性, 往往会造成人为的偏差, 使得局部异常加大或减小;趋势面拟合法, 使得原始数据点值发生了改变, 拟合的结果往往取得平滑的效果, 它可将长周期的趋势变化同短周期的局部变化分开;而趋势面加残差叠加法, 是将方向N－P法与趋势面拟合法有机的结合起来, 它克服了单独使用插值法或曲面拟合法的弊病, 产生了一个既能反映局部特征, 又能反映全局变化的网格数据。通过对比分析, 我们最终采用了趋势面加残差叠加法作为达里亚工区网格化的方法．实践证明, 利用该方法网格化后所形成的速度体完全适用于本工区的变速时深转换。

Four griddings are comparatively analyzed in this paper, they are N-P, N-Pwith orientation,trend surface fitting and trend surface stacking with residual error.It can be found through experiment that both N-P and N-P with orientation belong to the kind of interpolation method with which no original station measurements are changed and the computation is fast.However,N-P method does not take orientation into account, so it tends to lead artifical errors and makes local anomaly increase or decrcase. Trend surface fitting changes original station measurements and tends to give a smooth result, so we can distinguish local short term variation from long term trend.Trend surfacc stacking with residual error is the combination of N-P with orientation and trend surbo fitting, it Can overcome the shortcomings which occur for the application of interpolation or curved sarface fitting separately, and produces curved grid data that have bothlocal features and long term trend features. After comparative analpeis we have deckled to adopttrend surface stacking with residual error as the gridding method in the Daliya area. Our practices show that the velocity volumes produced by this method are fully applicable to the timedepth inversion of varying velocity in this area.