| The Geographical Information System (GIS) has become more and more commonly available over past 10 years, Meanwhile, interpolation of data sets is becoming a common problem for geographical and geological researchers. It is needed to build digital terrain models (DTM) in geochemical and geophysical exploration and the like.There exit abundant references on methods of interpolation for DTM and other models, with a number of them used as an entry point to the exploration of different interpolation models. Methods of interpolation may be divided into types: exact and approximate. The former preserves the values at data points, while the latter smoothes out the data. Another popular way to classify interpolation methods is by the influence range of the data points involved. Global and piecewise methods are applied for interpolation respectively when all sample points and only data nearby are taken into consideration. The surfaces obtained from conventional methods are usually non-stationary and non-periodic, the use of overly distant points may deform the interpolation surface. Therefore, a local and exact interpolation method on surface patches is widely considered satisfactory. With the development of nonlinear theories, interpolation methods are grouped into linear and non-linear ones. Non-linear interpolation is attracting more and more attention, especially those established on fractal thoery. It is considered that the graphs handled by fractal interpolation reveals more details, though there are still many problems to solve. The features of real topography and surface may be well reproduced by the fractal interpolation, their have the same spectrum with the real ones.ARC/INFO is a popular GIS system, and has powerful ability of spatial analysis. It would be a good idea to handle data using fractal interpolation and then display and analyze them using the ARC/INFO platform. This is also true in handling data from geographical survey and geological exploration.INTERFACE FOR ASC II FILES IN ARC/INFO In ARC/INFO, there are many interfaces for exchange data beween applications. In the TIN (Triangulation Irregular Network) model, for example, the surface model may be built from a wide variety of sources, including coverage. ASC II files such as xyz photogrametric data, pre-existing tins and sample points may be extracted from lattices. A corresponding command option is supplied in the Createtin command for different sources.The Generation option is used to input ASCII text files, containing x, y, z coordination for points, lines or polygons in the Generate input format. Different input files have different formats. The format for generate files containing points is :The feature ID is usually assigned a value of 1 in default. The end of the file is indicated with the end keyword.To convert coverages to generate files , the ARC/INFO supplies the ungenerate command. These commands and their options are used in this paper to exchange data with ARC/INFO.ALGORITHM OF SRA The Successive Random Addiction (SRA ) was, at first used to produce artificial topography. Its basis is the Brownian Motion and the fractal Brownian Motion (FBM). The following statement presents the main thought of SRA to produce a fractional Brownian curve (in 1 -dimension).A curve is produced on [0, 1] at the presumption that the values of y at the two ends are y0 and yl, respectively. So the value of yj/2 at the middle point (x = 1/2) may be gotten from the equation:in which, Ay is a perturbation motion, and Ay ~ N(0, x2H). As many points as need may be gotten through the same way.The idea can be developed to the 3-D irregular distribution point data. Supposed the value z at the point (x, y) be needed. The no points near the point (x, y) are selected according distance, and them are let different weight on distance, the average value z of these n0 points may be gotten. It is the basic value ZQ at (x, y). According to FBM, a perturbation motion Az ~ N(0, CT2(Ax2+Ay2)H) is added to the basic value ZQ, the final value z...
|
PDF Full Text Request