Digital terrain models derived from SRTM data and kriging

Abstract

The objective of this work is to define procedures to improve spatial resolution of SRTM data and to evaluate their applicability in the Serra Negra region, in the district of Patrocinio, state of Minas Gerais in Brazil. The region's structure is a result of past tectonic processes that have arched it into a dome shape. Besides the already existing agriculture exploitation, Serra Negra also has strong tourism and mining potential. The Digital Elevation Model (DEM) was done using different interpolation methods in a resolution of 30 meters or 1 arcsec, among them kriging, ideally used to manipulate random spatial variations due to its capacity for dealing with spatially variable components. The accuracy of the resulting DEM and slope maps modelled were evaluated based on slope measured in the field. The correlation coefficients were determined from the field data and those derived from the DEM. Analyses and tests with SRTM data, divulged for South America are presented to better adequate the model to the study area. The correlation coefficients in the estimates by kriging and by bicubic interpolator were similar, with a slight difference in favour of kriging. Therefore, kriging is an interesting alternative in elaborating Digital Elevation Models that are in keeping with the dome structure of the Serra Negra region. In order to measure operational aspects of the pre-processing methods, the study area data were prepared under a resolution of 30 meters and evaluated through statistical analysis and visualizations of the DEMs and slope. The data presented strong restrictions to being used in their original form due to the low spatial resolution. However, the pre-processing allows their use in relatively detailed scales. Based on the results, a proposal for the development of a DEM with the SRTM data for the Serra Negra region is presented.