Aeromagnetic Compensation with Partial Least Square Regression

Abstract

Magnetic exploration plays a significant role in regional geological investigation and detection of underground geological bodies with magnetic anomaly. At the moment, aeromagnetic survey is widely applied for its high efficiency, low cost and less subject to terrain restrict. Magnetic compensation is a key step in pre-processing survey data and several methods have been used. In this paper, we would use partial least square method to complete aeromagnetic compensation. Partial least square regression is frequently used to find the fundamental relations between two matrices. It combines linear regression analysis, canonical correlation analysis and principal components analysis. It can be applied into data with multicollinearity among independent variables and the number of variables is larger than that of observations. Before compensation, we should have several pre-processing steps such as parallax correction, diurnal variations correction, geomagnetic field correction and high-frequency noise removal. This will provide us magnetic data with higher quality and make compensation process more accurate. We set synthetic aeromagnetic data with interference of aircraft’s maneuvers and used partial least square method to do compensation. From the results of simulation, we can see that the interference signal is reduced to a low degree and satisfied compensation effect is obtained. Partial least square regression is a stable and effective method in the application of aeromagnetic compensation.