Jump to content

 
Photo

Creating contours from DTMs

- - - - -

  • Please log in to reply
2 replies to this topic

#1
Scotian Gold

Scotian Gold

    Newbie

  • Validated Member
  • Pip
  • 2 posts
  • Brazil

Hello to everyone! I work for a surveying and mapping company that is investing a fair amount of time and money into streamlining our processes. To this end, I have been asked to research methods for generating mapping-quality contours automatically from DTMs containing a regularly spaced grid, form lines and break lines.

Our operational reality is:

  • Our maps are produced at various scales between 1:1K and 1:20K, typically with 1m or 5m contour intervals.
  • Our DTMs will contain upwards of 50 million points, and implement form and break lines (hydrography, walls, manmade slopes, etc.)
  • Accuracy is more important than cartography, but we do want an 'attractively smoothed' accurate dataset that will require minimal editing. Otherwise, we will continue creating contours from stereopairs. Obviously we have more control here, but it is a lengthy process.
  • Map products are produced and delivered to our clients in DWG format.
Keeping this in mind, I have two questions:

  • What contouring software would you recommend?
  • What DTM cell size and smoothing would you use to generate n metre contours?
I believe that's all for now! I look forward to reading your responses!

Ron

#2
Lui

Lui

    Master Contributor

  • Validated Member
  • PipPipPipPip
  • 153 posts
  • Gender:Male
  • Location:Ljubljana
  • Interests:Natural looking maps, hillshading, mountain cartography, LIDAR, TLS, data fusion, photogrammetry, singing, poetry, cycling
  • Slovenia

Hello to everyone! I work for a surveying and mapping company that is investing a fair amount of time and money into streamlining our processes. To this end, I have been asked to research methods for generating mapping-quality contours automatically from DTMs containing a regularly spaced grid, form lines and break lines.

Our operational reality is:

  • Our maps are produced at various scales between 1:1K and 1:20K, typically with 1m or 5m contour intervals.
  • Our DTMs will contain upwards of 50 million points, and implement form and break lines (hydrography, walls, manmade slopes, etc.)
  • Accuracy is more important than cartography, but we do want an 'attractively smoothed' accurate dataset that will require minimal editing. Otherwise, we will continue creating contours from stereopairs. Obviously we have more control here, but it is a lengthy process.
  • Map products are produced and delivered to our clients in DWG format.
Keeping this in mind, I have two questions:

  • What contouring software would you recommend?
  • What DTM cell size and smoothing would you use to generate n metre contours?
I believe that's all for now! I look forward to reading your responses!

Ron


Hello Ron

I'm involved in large scale mapping (1:500 - 1:5000) with LIDAR point cloud as primary datasource and I have recently came up with rather unique "contouring" approach that is implemented in latest TerraScan and Terramodeler LIDAR software. How it goes?
First: classification of ground points has to be done (automatic and manual) point cloud -> ground points
Second: from point cloud breaklines (hard and soft) are extracted (mostly manually) then significant areal (plane) relief features are 3D vectorised (roads, building footprints, embankments,...)
Third: ground points are classified in separate countour point cloud using unique approach in TerraScan that enable us to set how "nice" aka. smooth contours should be (from accurate to smooth). Algorithm select ground points that portraied the best result. Algorithm also classify to contour points significant relief point as peaks and pits where smallest area of each can be set. This approach is opposite from what we are used to in common GIS software. Usually point heights are " hard baked" in contour creation process.
Fourth: early vectorised features (relief breaklines, planes) and contour points are triangulated using so called rules. Rules describe what kind of influence have vectorised features in contour creation like buildings or roads can be holes or planar relief objects,... In this process characteristic relief points (peaks and pits) with height label are also produced.
Resulting contours are (when right accuracy - smooth factor is set) very good and I'm using them as they are produced.
Well I know that if you are not processing LIDAR datasets then using mentioned software is rather expensive and unlogical but it use different approach to contour making and I just have to mention it.

Lui

#3
Scotian Gold

Scotian Gold

    Newbie

  • Validated Member
  • Pip
  • 2 posts
  • Brazil

Hello Lui, thanks for your detailed response! We are using Inpho’s ApplicationsMaster 5.1 (Match-AT, Match-T and DTMaster) to generate DSMs and DTMs. Point clouds can be generated with densities similar to that of LIDAR data. Break and form lines are currently captured from stereopairs, but as Match-T and DTMaster are our most recent acquisitions, I have not seen if they have automatic break and form line extraction utilities.

Do you resample your point cloud based on your contour interval or the scale of your map? One of my challenges is to minimize the size of the point cloud; I want to determine the largest DTM cell size possible that allows for the creation of accurate contours. For this, I consider two types of cell size: a standard paper pixel size and a maximum ground pixel size.

If I can determine a standard paper pixel size, then ground pixel size may be irrelevant. By creating 1.0 mm DTMs at any map scale, the density of the grid will always remain the same. This should therefore create relatively consistent looking contours across maps and scales, when the ratio between contour interval and map scale is identical.

However, I believe it will be important to determine a maximum ground pixel size which will override the standard paper pixel size if less than 1.0 mm at map scale. This value could be determined by the spacing and density of form and break lines, to guarantee that there are at least n number of points between these features to correctly interpolate between them. If anyone has information on what this n value should be, I’d like to see it!

Ron




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->