Skip to the main content

Original scientific paper

https://doi.org/10.17559/TV-20131010210942

The accuracy of deterministic models of interpolation in the process of generating a digital terrain model - the example of the Vrana Lake Nature Park

Ante Šiljeg orcid id orcid.org/0000-0001-6332-174X ; Sveučilište u Zadru, Odjel za geografiju, Trg kneza Višeslava 9, 2300 Zadar, Croatia
Sanja Lozić ; Sveučilište u Zadru, Odjel za geografiju, Trg kneza Višeslava 9, 2300 Zadar, Croatia
Silvija Šiljeg ; Sveučilište u Zadru, Odjel za geografiju, Trg kneza Višeslava 9, 2300 Zadar, Croatia


Full text: croatian pdf 2.354 Kb

page 853-863

downloads: 703

cite

Full text: english pdf 2.354 Kb

page 853-863

downloads: 682

cite


Abstract

The aim of this research is to choose the most appropriate interpolators for the purpose of geomorphologic research, by comparing 7 deterministic methods. For the purpose of developing a model and comparing interpolators, the research used a set of elevation data gathered by aero-photogrammetry and stereo-restitution. The accuracy of interpolation methods was tested by analysing 8 statistic parameters, which were obtained by methods of cross-validation, split sampling and jackknifing. Apart from analysing the parameters, the research also compared the interpolation methods by visual means, through graphic representation of data (two-dimensional and three-dimensional) on the basis of credible graphic representations of sets of data. The research also tested the effect of power, number of neighbours, distance and sectors on the quality of output data. The multiquadratic radial basis function (MQ-RBF) proved to be the best deterministic method of interpolation by all the relevant parameters.

Keywords

digital terrain model; deterministic interpolation methods; methods of comparison of interpolation methods; root mean square error; Vrana Lake Nature Park

Hrčak ID:

143096

URI

https://hrcak.srce.hr/143096

Publication date:

8.8.2015.

Article data in other languages: croatian

Visits: 2.662 *