Comparison Of Geochemical Data Interpolation Methods

A geochemical survey work, often collect dozens, hundreds or even thousands ofsamples, each sample analyze a dozen or even dozens of elements (or other chemicalindicators), access to thousands of millions of various types of data. How to obtain usefulinformation from a number of confused the data, identification of noise, focused on thevital grasp the "subject" remove "branches", need for the acquired information to thescientific processing order. In the course of geochemical data processing, the data need forinterpolation. Interpolation is an important method of discrete function approximation,using it which can through the function value condition in a finite number of points toestimate the approximationthe of the function at other points. Better interpolation methodcan not only effectively simulate the spatial distribution of geochemical variables, and onthe basis of retain the original effectively information can be more reasonable to distinctbackground and anomalies of geochemical variables, having important effects ofdiscoverring and strengthening weak anomalous information, so as to provide a reliablebasis to determine targets.The normal data of exploration geochemistry are often follow normal or lognormaldistribution, but the anomaly data are often follow fractal distribution and local singularity.In this paper, quantitative methods compare with interpolation results of radial basisfunction interpolation, fractal interpolation, inverse distance weighted interpolation,Kriging interpolation in geochemical data, for a reasonable choice of interpolation methodattempting to provide a quantitative basis.Inverse distance weighted method is a typical representative of the geometric method,which is mainly based on distances between the data to interpolate; Radial basis functioninterpolation method is a function interpolation, and it interpolates from the surface must bestudied within each known sample point, which is one of artificial neural network. Kriginginterpolation is part of geostatistic interpolations, which is based on spatial autocorrelation,through the correlation of data spatial structure to interpolate. Fractal interpolation is basedon principles of self-similarity, according to the scale invariance, through iterated functionsystems to achieve its interpolation. When set for the same data and model parameters,analysis and use the differents from the models among of Inverse Distance Weightedinterpolation, Radial Basis Function interpolation, Kriging interpolation, Fractalinterpolation. In this paper, author selected four different data, in order to study fourinterpolation methods application on geochemical data, the four sets of data are: the data subject to fractal distribution, the data subject to the Standard normal distribution, the datasubject to the Standard lognormal distribution, the actual data come from Chagan Mulongregion of Chifeng City of Inner Mongolia Autonomous Region. For these four differentdata, author analyzes the results of difference between the original data and theinterpolation results from different angles.