Skaitmeninio reljefo modelio sudarymas skirtingais geostatistiniais reljefo modeliavimo metodais
Abstract
The polynomial method usually employs the regression model created by all data. A response surface is fitted to the x and y coordinates which are the covariates. Polynomial interpolation uses data within a localized territory rather than all of the data, so it fits local trends and uses weights. To carry out the experiment, the Vilnius city territory in Šeškinės Ozas was chosen. The territory served for obtaining detailed and high accuracy initial data by the stereophotogrammetric method (the standard deviation of altitude points is 0, 14 m, the average distance between the points being about 6 m). High points should be sufficient dense, and they were enough not only for the terrain model, but also for the monitoring accuracy. The objective of the current work was to investigate the accuracy dependence of digital terrain models on the methods of initial data preparation and modeling. With a view to create a powerful model, was explored the influence of the accuracy of the modeling methods. The best results were secured when DTM was created by the kriging method using a circular semivariogram and by interpolation taking 12 neighbouring high points.