Study For Scaling Of Data Events

Multi-point geostatistical simulation algorithm based on pattern is a common geological modeling algorithm,but these algorithms are faced with the problem of data event template optimization.If the template scale is too small,the template cannot recognize the large spatial structure of the training image,while if the template scale is too large,the data event dimension is too high,and the calculation requires too much space and storage resources,which is quite time-consuming.At present,the general template optimization method is to calculate the two-point information entropy of the training image at each template scale.The information entropy of the training image can be used to describe the uncertainty of each template assignment to the simulation grid.The larger the entropy value is,the greater the uncertainty is,and the smaller the entropy value is,the smaller the uncertainty is.Therefore,the optimal template scale can be determined by finding the minimum template scale that can make the training image reach the maximum entropy.In this paper,the optimal template selection of stationary training images is studied theoretically and empirically.By establishing several mathematical models for the information entropy of training images at different template scales,it is found that the exponential model can fit the scatter diagram of the information entropy well,and it is concluded that three times of the scale parameters of the exponential model is the critical scale of information,so the exponential model is selected as the information entropy model.After calculation,it is found that the information entropy model allows the information critical scale to be obtained quickly by calculating the information entropy of the first and last two template scales of simple training images,and only 35% of the information entropy template scales of all templates are needed to calculate the information critical scale of complex training images.In this paper,the information entropy model is coupled with the multi-resolution simulation algorithm,and it is found that the critical scale of information will scale equally with the scaling of the training image,which can describe the scale effect of data events.The accuracy of multiresolution simulation algorithm is improved under the same algorithm complexity.