Abstract:Interpreting gravimetric and magnetic data by random process theory brings unique effect because it can well overcome the difficulty conventional methods can hardly do. It is proposed that the random process can be expressed by autocorrelation function. The transformation of random gravity anomaly into autocorrelation function also brings upward continuation effect of gravity anomaly, so that the autocorrelation function can remove high-frequency noise and observation error, and enhance the major anomaly caused by gravity-field source. Tlhe autocorrelation functions of random gravity anomaly can be classed into live types, namely, unit impulsive, triangular, Markovian, Gaussian and cosine-exponential types. Using the models of these types, we can obtain corresponding normalised autocorrelation function, then derive both the shape functions of corresponding geological bodies and the estimation functions for calculating center depths of the geological bodies. Two real gravity profiles were processed by using above theory, then the processed results were compared with seismic and geological profiles. The comparison shows that using the estimation function to calculate average depth of major density layer and to constr ct geological-geophysical model for forward and inverse calculations favours analysing the formation classification and bed-surface undulation. It must be particularly pointed out that the average depth of major density layer is derived from gravimetric data, without the dependence of density data.