Study On Statistical Inference Methods Of Three Dimensional Geological Features Between Boreholes

Geological features are hosted in certain geological backgrounds, their spatialdistributions under different geological backgrounds have different characteristics.For example, topographic relief (one of the geological features) seems to be veryintense in mountainous and weathering areas while it is relatively flat in plains andsedimentary areas. In order to reflect the random and structural characteristics of thegeological features accurately, we should fully consider the objective relationshipbetween geological features and geological backgrounds. However, most of thespatial interpolation methods are pure mathematics presently, to the extent that thebuilding precision of "3D Strata Information Systems3DSIS" is low, which has agreat impact on the realization of "digital earth" process.Basing on the geological random field theory and spatial interpolation methods,using surfer utility software and IDL programming language as tools and taking theactual projects as a support, the thesis devote to study the3D statistical inferencemethods of geological features between boreholes. The main research work andconclusions are as follows:1. The basic principles of the10kinds of spatial interpolation methods wereanalyzed and compared, the characteristics of each method, the scopes of applicationand the existing problems were discussed, and the thesis obtained some certainunderstandings. Among them, the Kriging method is a method which can well reflectthe random characteristics of geological features, it is worth to recommend; Theinverse distance weighted(IDW) is simple in principle and flexible in weightfunction，although it does not consider the influence of the geological backgroundspresently, but it is feasible in theory and can be a good interpolation methodconsidering the geological backgrounds by determining the interpolationneighborhoods and weights properly.2. Two typically projects——Zhuhai Doumen District project(delta plains) andGuangzhou Baiyun District project(low mountains and hills) were selected toresearch the correlation of geological features under different geological backgrounds,the thesis calculated and analyzed the correlation distances,correlation functionsand variogram reduction functions of the soil layers and the ground water. The calculatingresults can be use as an important parameter to determine the neighborhoods andweights of interpolation which apply to the spatial interpolation in corresponding soil layers and ground water,it can also be taken as reference in projects under similargeological backgrounds;3. Due to the big error caused by the limitation of the number and distribution ofboreholes, the variogram is hard to be modeled. The thesis discussed the analyticalmodels of variogram,standard types of correlation function and theirparameters’relationship,deduced a method which providing a new way to determinethe variation model of Kriging method by using correlation distance (or correlationfunction) to determine the parameters of variogram models.4. Bringing in the correlation distance to determine the neighborhoods andweights of interpolation, a new statistical inference method—Correlation WeightedInterpolation (CWI) were presented, this method fully considered the relevantcharacteristics of geological features and overcomes the subjectivity whendetermining the power p in the IDW method.The CWI is a powerful3D statisticalinference method of geological features between boreholes basing on geologicalbackgrounds.