基于差分求导的伪总梯度余弦值法的位场边界检测

Potential field edge detection based on difference derivative with cosine of pseudo total gradient

针对利用位场资料解决位场边界的精细划分和解释问题,本文提出了一种新的位场边界识别方法——基于差分求导的伪总梯度余弦值法。利用窗口滑动模式,对异常平均值采用差分方法计算导数,并引入伪总梯度的概念,以伪总梯度与观测面夹角的余弦值作为滤波输出,计算过程中对边界位置的极值特征进行了两次紧缩,其最大值与地质体边界相对应。模型实验对比结果表明,该方法稳定性较强,对边界平面位置的识别效果更好。在实际数据处理中,推测出黑龙江省虎林盆地的29条断裂,其中12条断裂的位置与欧拉反褶积反演结果一致,并有7条断裂在以往的研究中得到证实,证明了本文方法对地质体边界平面位置的识别能力。

Aimed at edge refined division and interpretation of potential field,we propose in the paper a new edge detection approach based on difference derivative with cosine of pseudo total gradient.We first calculate derivatives of the mean of anomaly with difference method in sliding windows.Then we import a concept of pseudo total gradient,and take the cosine of angle between pseudo total gradient and observation plane as the filter output.In the process,the maximum features on the edge position are tightened for two times,and its maximum corresponds to the geological edge.Model tests show that this method is more stable,and can better identify geological edge plane than other methods.In a pilot application,this method de-tects 29 faults in Hulin Basin,Heilongjiang,from which 12 faults are consistent with the result of Euler deconvolution and 7faults were proved in the previous research.