New technique for selecting the vertices for a TIN and a comparison of TINs and DEMs over a variety of surfaces

Mark P. Kumler, Michael Goodchild

Research output: Contribution to conferencePaper

Abstract

A technique is presented for selecting the vertices for a Triangulated Irregular Network (TIN) from digitized contour lines and selected evalation points. The procedure relies on a Douglas-Peucker generalization of the contour lines to capture a large set of points along ridges and channels. This set is supplemented by elevation values at peaks and pits, along edges and at corners of the model, and at other locations necessary to improve the overall fit of the TIN. The procedures for identifying the supplementary points are described in detail. In addition, a comparison is presented between contour-based TINs and conventional, gridded DEMs for 11 study areas from around the United States, including one from each of the major physical divisions described by Fenneman. TINs are created for each study area that require the same amount of disk space as the corresponding DEMs, by limiting the number of vertices to approximately one-tenth the number of elevation values in the DEMs. The two structures are compared by determining how well they estimate the elevations at over one thousand test points, and by visual inspection of perspective block diagrams and sharled relief maps. The contour-based TINs yield superior images and consistently lower root mean square errors.

Original languageEnglish (US)
Number of pages1
StatePublished - Jan 1 1991
Externally publishedYes

Fingerprint

Mean square error
Inspection

ASJC Scopus subject areas

  • Engineering(all)

Cite this

@conference{ec2f88ea46934fa280bef8398c54de4b,
title = "New technique for selecting the vertices for a TIN and a comparison of TINs and DEMs over a variety of surfaces",
abstract = "A technique is presented for selecting the vertices for a Triangulated Irregular Network (TIN) from digitized contour lines and selected evalation points. The procedure relies on a Douglas-Peucker generalization of the contour lines to capture a large set of points along ridges and channels. This set is supplemented by elevation values at peaks and pits, along edges and at corners of the model, and at other locations necessary to improve the overall fit of the TIN. The procedures for identifying the supplementary points are described in detail. In addition, a comparison is presented between contour-based TINs and conventional, gridded DEMs for 11 study areas from around the United States, including one from each of the major physical divisions described by Fenneman. TINs are created for each study area that require the same amount of disk space as the corresponding DEMs, by limiting the number of vertices to approximately one-tenth the number of elevation values in the DEMs. The two structures are compared by determining how well they estimate the elevations at over one thousand test points, and by visual inspection of perspective block diagrams and sharled relief maps. The contour-based TINs yield superior images and consistently lower root mean square errors.",
author = "Kumler, {Mark P.} and Michael Goodchild",
year = "1991",
month = "1",
day = "1",
language = "English (US)",

}

TY - CONF

T1 - New technique for selecting the vertices for a TIN and a comparison of TINs and DEMs over a variety of surfaces

AU - Kumler, Mark P.

AU - Goodchild, Michael

PY - 1991/1/1

Y1 - 1991/1/1

N2 - A technique is presented for selecting the vertices for a Triangulated Irregular Network (TIN) from digitized contour lines and selected evalation points. The procedure relies on a Douglas-Peucker generalization of the contour lines to capture a large set of points along ridges and channels. This set is supplemented by elevation values at peaks and pits, along edges and at corners of the model, and at other locations necessary to improve the overall fit of the TIN. The procedures for identifying the supplementary points are described in detail. In addition, a comparison is presented between contour-based TINs and conventional, gridded DEMs for 11 study areas from around the United States, including one from each of the major physical divisions described by Fenneman. TINs are created for each study area that require the same amount of disk space as the corresponding DEMs, by limiting the number of vertices to approximately one-tenth the number of elevation values in the DEMs. The two structures are compared by determining how well they estimate the elevations at over one thousand test points, and by visual inspection of perspective block diagrams and sharled relief maps. The contour-based TINs yield superior images and consistently lower root mean square errors.

AB - A technique is presented for selecting the vertices for a Triangulated Irregular Network (TIN) from digitized contour lines and selected evalation points. The procedure relies on a Douglas-Peucker generalization of the contour lines to capture a large set of points along ridges and channels. This set is supplemented by elevation values at peaks and pits, along edges and at corners of the model, and at other locations necessary to improve the overall fit of the TIN. The procedures for identifying the supplementary points are described in detail. In addition, a comparison is presented between contour-based TINs and conventional, gridded DEMs for 11 study areas from around the United States, including one from each of the major physical divisions described by Fenneman. TINs are created for each study area that require the same amount of disk space as the corresponding DEMs, by limiting the number of vertices to approximately one-tenth the number of elevation values in the DEMs. The two structures are compared by determining how well they estimate the elevations at over one thousand test points, and by visual inspection of perspective block diagrams and sharled relief maps. The contour-based TINs yield superior images and consistently lower root mean square errors.

UR - http://www.scopus.com/inward/record.url?scp=0025754515&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0025754515&partnerID=8YFLogxK

M3 - Paper

ER -