Slashgeo

DEM Slope Algorithms and Irregular Grids

posted by Satri on Friday February 23, @11:21AM        Permalink  Trackback URI  Slashdotthis  Diggthis  Del.icio.us
from the slowly-sliding-to-the-bottom-of-the-DEM dept.

I took some time to look at DEM slope algorithms in order to integrate the code to our in-house open source geo-application. I found very interesting information, including this article named "A comparison of algorithms used to compute hill slope as a property of the DEM" by KH Jones [pdf]. This second interesting related article by Jones is named "A Comparison of Two Approaches to Ranking Algorithms Used to Compute Hill Slopes" [pdf]). This allowed me to learn that the Horn algorithm used in the open source GRASS GIS is rather very good, the other good algorithm is Fleming and Hoffer's. In the first article, Fleming and Hoffer's algorithm is found better for 'smooth surfaces' (read article to learn the context) than Horn's algorithm, while from the second article's abstract: "Horn's method (used in ArcInfo GRID) performed better than Fleming and Hoffer's as a slope estimator when the noise amplitude was very much larger than the MSED [mean smooth elevation difference].". Here are my questions: (a) does anyone know if Fleming and Hoffer's algorithm code can be found somewhere? (b) Does anyone knows DEM slope algorithms that can efficiently (computationally speaking) handle slightly irregular grids? Thanks!

Trackbacks:

• NONE
• NONE
• NONE
• NONE
• NONE
• NONE
• NONE
• NONE
• NONE
• NONE

Related Stories

An Inventory of SRTM-DEM Versions 7 comments [+]

I invite everyone to comment and improve this tentative to gather relevant information on SRTM-DEM data and versions. The objective is to build an inventory of the different SRTM-DEM versions. Versions considered are those which (1) do not degrade the spatial resolution of the original dataset and (2) which maintain the original near-global coverage. Read more below.

• Slope research is fine but we need better DEM's

  (Score:2, Insightful)

  by Green Palolo (918) on Friday February 23, @02:24PM (#1251)

  I think it is good to improve algorithms for elevation processing, however, without quality data a quality answer cannot be produced. The USGS 10 meter DEM's have artifacts that yield poor answers. The contour biasing is the worst problem when it comes to slope calculations. This artifact makes the slope look like a series of terraces, flat above the contour line and steep downslope.
  David Hatfield has a good article (December 1999) at: http://www.gisvisionmag.com/vision.php?article=%2F GISVision%2FTechPaper%2FTopoTools.html [gisvisionmag.com]
  • Re:Slope research is fine but we need better DEM's

    (Score:3)

    by Satri (3) on Friday February 23, @03:39PM (#1255)
    ( http://alexandreleroux.blogspot.com/ | Last Journal: Friday March 17, @04:07PM )

    Thanks for the feedback. The importance of DEM data quality somewhat depends on what you plan to do the data, in my case, I often work with meteorological models that use, at this moment, 900m orography from the GTOPO30 dataset. Downscaling SRTM-DEM near-global data would provide a much higher DEM resolution even if ultimately the data would be used at, say, 250m.

    [ Parent ]
  • USDA's ENFORCE

    (Score:3)

    by Satri (3) on Friday February 23, @03:50PM (#1256)
    ( http://alexandreleroux.blogspot.com/ | Last Journal: Friday March 17, @04:07PM )

    I read the article and found it quite interesting :-) I even thought USDA's ENFORCE software could be quite useful to improve DEM generated from contour lines, however, they sell it at 500$US...

    [ Parent ]
• references

  (Score:2, Informative)

  by Anonymous Voxel on Friday February 23, @02:58PM (#1253)

  The following reference might be useful:
  
  Meyer T.H ., Eriksson, M., and Maggio, R. C. (2001) Gradient Estimation from Irregularly Spaced Data Sets. In Mathematical Geology , vol. 33, No. 6, pp. 693-717.
  
          EBo --
  
  • Re:references

    (Score:3)

    by Satri (3) on Monday February 26, @01:53PM (#1264)
    ( http://alexandreleroux.blogspot.com/ | Last Journal: Friday March 17, @04:07PM )

    Thanks. Useful. There are two things that I wonder: (1) is there public domain algorithms and code to use this approach and (2) is this computationally efficient? (I fear computing DEM slopes of irregular grids that way could be very slow)

    [ Parent ]
• Irregular grids: here's what we done

  (Score:2)

  by Satri (3) on Thursday March 15, @08:51AM (#1315)
  ( http://alexandreleroux.blogspot.com/ | Last Journal: Friday March 17, @04:07PM )

  To solve the irregular grids problem, we simply modified GRASS GIS's implementation of the Horn algorithm to measure distance between each pair of points. From the tests we did, the results looks good.