29 replies to this topic

[GIS in the Annapurna Range]

Hi,

I have been working on trying to create a high resolution DEM for the Annapurna Range in Central Nepal. In this area SRTM data has voids because of steep terrain. DEM data for Nepal is not readily available from the public, unless you want to pay a large sum of money per DEM region.

To avoid cost, we have spent the last year digitizing scanned topographic maps for Central Nepal using ArcGIS 9.1. We have finally finished correcting errors caused by the autotrace program we used. I am now trying to make a high resolution DEM from our contour lines.

Interpolation methods in ArcGIS are used mostly for point, not line data. The topo to raster tool does not work for my data because it says there are too many points on the line. I could make the tool work by deleting points, but then our data would not have the detail we need.

We need to have a high resolution DEM because we are planning to look at the variations of steepness and curvature within individual rock units.

I have tried making a TIN from the contours using ArcGIS, but I get a terracing effect. This website outlines this problem pretty well: http://gis.esri.com/...pap924/p924.htm

Creating a TIN from the contours, then using the TIN to create a DEM incorporates these regions of flat slope. This is a huge problem since I want to extract and use accurate steepness and curvature values. I have read that some people have developed programs to mitigate this problem, but I have not been able to find a program that I could figure out how to operate yet.

Does any one have any advice to help solve my problem or know of a program I can use to generate a DEM from contour lines that are in a shapefile format.

Any help is much appreciated! Thanks!

[Hans van der Maarel]

Have you tried other methods than Arc for the gridding?

If you can send me your data, I'd be happy to give it a try in Global Mapper and VNS's terrain gridder.

In addition to that, I was wondering whether gap-filled SRTM would be suitable for your needs. Either the CGIAR dataset or the SRTMFill utility, which I personally prefer.

[François Goulet]

Have you tried other methods than Arc for the gridding?

With PCI Geomatica, it's a 3-clicks operation. I could try if for you.

[Charles Syrett]

I've never used Arc, but I've tried Manifold, Surfer, and Global Mapper in making DEMs from contours. By far the best performance I got was from Manifold, using the DEST algorithm. The terracing that plagues other methods is almost nonexistent, and if you create new contours from the DEM they'll be very close to the original input contours.

Charles Syrett
Map Graphics
http://www.mapgraphics.com

[ProMapper]

I have not used Manifold, so can not comment on its efficacy. However I have used MTA (Modular Terrain Analyst), an MGE application which runs in Microstation environment. It generated DEM without any terracing effect.

If you want to minimize terracing effect in Global Mapper then choose the height unit as milimeter rather than meter. I keep using Global Mapper for exporting DEMs and found this technique which does not generate terracing. May be you can try reconverting through Global Mapper.

However I would be greatly interested in the work flow for contour digitisation. If it is not a big hassle could you share that with the users of Cartotalk.

Thanks

Anu
http://www.mapsandlocations.com

[Kevin McManigal]

Maybe you could clip the contours with a polygon to get the node count down without losing detail. Then use the Topo to DEM tool, recombining the DEMs with a mosaic operation in the raster calculator. This could all be done in Model Builder to speed up the repetitive process.

Good luck, I'd love to see the final maps!

[Jimbo]

I have not used Manifold, so can not comment on its efficacy.

If you have a NVIDIA capable of being a super computer chip then what would take hours to process, (especially if your processing a very high quality DEM with a lot of contours such as the data from GeoNova), then it could be mere seconds. But we have to consider the smaller the file size, the shorter amount of time & the worse quality it comes out to be.

[rudy]

I've never used Arc, but I've tried Manifold, Surfer, and Global Mapper in making DEMs from contours. By far the best performance I got was from Manifold, using the DEST algorithm. The terracing that plagues other methods is almost nonexistent, and if you create new contours from the DEM they'll be very close to the original input contours.

Charles Syrett
Map Graphics
http://www.mapgraphics.com

Charles -

I've been trying to generate a useable DEM from a contour sample and have run into the same problem of the stepped appearance mentioned earlier. I've been using Manifold but am not aware of the DEST algorithm. Can you elaborate? I've got the basic version of Manifold 8 . . . am I going to need something more?

[Jimbo]

I've never used Arc, but I've tried Manifold, Surfer, and Global Mapper in making DEMs from contours. By far the best performance I got was from Manifold, using the DEST algorithm. The terracing that plagues other methods is almost nonexistent, and if you create new contours from the DEM they'll be very close to the original input contours.

Charles Syrett
Map Graphics
http://www.mapgraphics.com

Charles -

I've been trying to generate a useable DEM from a contour sample and have run into the same problem of the stepped appearance mentioned earlier. I've been using Manifold but am not aware of the DEST algorithm. Can you elaborate? I've got the basic version of Manifold 8 . . . am I going to need something more?

There's your problem rate there, your license limits the DEM to the default "Kriging" option. What Charles is referring to is the option to adjust your contours using the triangulation method which is an option within the Manifold Surface Tools + Personnel Edition license.

So just to show you an example of DEM processed with the DEST method:
 Kings_CT.png   232.58K   185 downloads

[Michael Schmeling]

GRASS should be able to do this with the r.surf.contour command.

And GMT also has several methods for generating grids, e.g. the surface command, which makes a grid from randomly-spaced (x,y,z) triples. May be worth trying these out.

[rudy]

So just to show you an example of DEM processed with the DEST method:
 Kings_CT.png   232.58K   185 downloads

Nice looking DEM . . . . I'll have to consider an upgrade. Can the Surface Tools run on just the personal edition or am I going to need to get the enterprise edition as well?

[Jimbo]

So just to show you an example of DEM processed with the DEST method:
 Kings_CT.png   232.58K   185 downloads

Nice looking DEM . . . . I'll have to consider an upgrade. Can the Surface Tools run on just the personal edition or am I going to need to get the enterprise edition as well?

Surface tools can run a Personnel edition if you have $145 US dollars to spend, however I would chose to spend $225 USD on all of the extensions which includes your surface, geocoding, & business tools.

[Clark Geomatics]

With Manifold, the incremental costs are always worth it - you get so much bang for your $. Remember to go for a 64-bit OS if you are a serious Manifold user, the results are worth it. As per Charles' reply, Manifold's implementation of the DEST algorithm will give you the best results when creating DEMs from elevation isolines.

[rudy]

With Manifold, the incremental costs are always worth it - you get so much bang for your $. Remember to go for a 64-bit OS if you are a serious Manifold user, the results are worth it. As per Charles' reply, Manifold's implementation of the DEST algorithm will give you the best results when creating DEMs from elevation isolines.

So how is Manifold's speed at producing DEMs? In the personal edition it can take from an hour and and half to more, depending on the parameters set, to produce at 1200 x 1200 dpi DEM.

[Jimbo]

With Manifold, the incremental costs are always worth it - you get so much bang for your $. Remember to go for a 64-bit OS if you are a serious Manifold user, the results are worth it. As per Charles' reply, Manifold's implementation of the DEST algorithm will give you the best results when creating DEMs from elevation isolines.

So how is Manifold's speed at producing DEMs? In the personal edition it can take from an hour and and half to more, depending on the parameters set, to produce at 1200 x 1200 dpi DEM.

The sample I shown you was with 32 bit at 5 metre contours; it took 2.5 minutes. If you have a NVIDIA CUDA 8100 or newer graphics chip it would have taken seconds.