位场归一化差分法的边界检测技术

针对常规位场数据处理存在场源边界定位精度低和识别能力差的问题，提出了识别边界位置的归一化差分法。根据x、y、z 3个方向的差分算子与构造边界位置位场异常特征间的关系，给出了突出异常梯级带的n阶归一化差分表达式。单体模型试验表明：z方向和经90°相移的x、y方向的一阶差分可以对模型体边界位置的异常梯级带进行紧缩；一阶归一化差分的异常等值线较为集中地分布在模型体边界位置，说明对边界有一定的识别能力，但精度稍低。为此，采用二阶差分进行进一步试算。结果表明:3个方向的二阶差分在一阶差分的基础上进一步增强了对边界的识别能力，而且二阶归一化差分的异常梯度带与模型边界位置具有良好的一致性。含噪声组合模型重力异常对比分析表明：与常规边界识别方法相比，差分半径较小的二阶归一化差分对地质体的边界识别能力强、定位精度高；差分半径较大的二阶归一化差分在一定程度上可以解决噪声干扰的问题，且等值线梯级带可以相对地突出大型地质体的边界。在实例应用中，归一化差分法检测出了黑龙江虎林盆地28条断裂，其中13条断裂通过盆地已有单元构造格局和DB1线电阻率剖面得到了证实。

Edge Detection of Potential Field Using Normalized Differential

Due to poor positioning precision and weak recognition capability of structur edge in conventional potential-field data processing, we present normalized differential method. According to the relationship between three-directional differential and the character of potential-field anomaly in position of structured edges, we give expressions of n-order normalized differential which can protrude the character of anomaly gradient zones. It is shown in a test of single body model that x-and y-directional first order differential after 90° phase shift and z-directional first-order differential can compress the width of anomaly gradient zones, and the contours of anomaly gradient zones center coincide the model edges. In order to improve positioning precision, we adopt second-order differential to pilot calculation. The results show that three-directional second-order differential further enhance edge recognition capability, and reveal the true model bodies. Thus it clarifies that the second-order normalized differential has higher positioning precision in edge detecting. Numerical test with noisy data shows that second-order normalized differential has strong edge recognition capability and high positioning precision when the difference radius is small, while normalized differential can effectively reduce the noise effect on data with larger geologic bodies’ boundary when the differential radius is large. In an application, normalized differential method detects 28 faults in Hulin basin of Heilongjiang, and 13 faults confirmed by geological survey on profile line DB1.