On the errors of spatial visualization of digital data in gravimetry

Purpose of thesis is to present some notes on guided interpolation for optimal digitization of maps. When visualizing the large volumes of geophysical data, the interpreter encounters the anomaly identification problem. The source of the ambiguity of the formal visualization of the data is its reduction onto a regular network. Large amounts of data, interpolated “by default”, give the unreliable distributions. The image of useful signal is distorted when the gravity field is poorly differentiated or has anomalies of a gully type. As a result, the array of interpolated data differs from the real data distribution. Further transformations of an array will distort the geological content of the data.

Approach proposed is to choose properly an interpolation technique and to adjust its parameters truly. A 3D surface with minimal distortion relative to the original map is obtained by kriging and modified Shepard's methods. But pairwise comparison of gridding techn iques on simulated gravity datasets, based on the minimum degree of deviation of the contour maps generated showed another output.

Results of the application of 12 interpolation techniques in Golden Software Surfer to the gravity data and their linear transformants are studied. The visualization of real gravity data with a gradient distribution of anomalies is ambiguous, especially for the noised data. Its quality depends on the data density, network geometry, the level of survey errors, and the nature of the anomalies. The image closest to the real field was obtained by the techniques of minimal curvature and radial basis function.

Practical value of the fine tuning of the interpolation procedure is proved. The data visualization should be carried out using guided interpolation, including the calculation of interpolation errors and derivatives of the gravity analytical model.