3

I have a 5 meter interval contour lines feature class which I would like to interpolate to a DEM.

I am well aware of at least 2 tools ("tin to raster" and "topo to raster") which can help me with that. To produce the proper DEM I need to define first its purpose.

Long story short - I need to track the progression\advancement of small streams (AKA: gullies). They are typically 1 meter wide and 2 m long. My algorithm which will be using the DEM as a basis, must have as much as accurate DEM as possible. How accurate? - detailed enough to see an advancement of a 30 cm length for example.

To do that I must interpolate (from the mentioned above contours) a most pleasing DEM.

The questions is: what course of action would you suggest me to implement? How can I choose the correct parameters?

I know trial and error would be the most advisable answer, yet there are so many parameters (smoothing, cell size, method type, enforce drainage, tolerance) to consider it might take me weeks until I get the correct DEM.

  • 2
    Welcome to the GIS Stack Exchange. If you could, tell us what software are you using by including it in your question and/or tags. – Barbarossa Nov 26 '13 at 15:09
  • It seems you are using ArcGIS. Try-and-error won't take weeks. Try the smoothing parameters: the discretization error factor (1.5, 2 or higher), tolerance #1 (try 2-3) and tolerance #2 (about 100), and on-off drainage enforcement. – nadya Nov 26 '13 at 20:16
  • First of all, thank you for responding so quick. Nadya, what you argue is true to some degree. but think of it as a factorial calculation. With every change that i make for one of those parameters the other could remain the same or be adjust accordingly. So relating to your mentioned parameters i could have a least 50 options or so without considering actual other parameters and without considersing that those paramaters have a range of values. – user24379 Nov 27 '13 at 13:13
  • You have not yet mentioned what software you are using, which makes this question unanswerable for now. Although I have to comment on the fact that you are using 5 m equidistant contour lines and want details down to 3 dm in height - which IMO is not possible. Even 0.5 m contour lines would not be likely to accomplish such accuracy. You need the point cloud (laser scanning) that the contour lines are likely to have been derived from, and create your DEM from that directly (your accuracy then depends on the accuracy of the points). – Martin Nov 28 '13 at 8:32
  • Depending on how many points per sqm you have (typically 3-4), you might be able to track structures at a 1-2 meter size in plane. Accuracy in height is depending on the method and what surfaces you measure - hard surfaces (eg. rooftops etc) have higher accuracy than grass fields, areas with bushes and lowest where there are trees present. Also, water surfaces are usually not reliably measured. – Martin Nov 28 '13 at 8:38
3

For any surface analysis involving hydrology, TopoToRaster is preferred over TIN as it has Drainage Enforcement.

As @Martin noted, 5m contour interval is unlikely to be dense enough to create or track sub-meter features.

That said the vertical contour interval is less important than the distance between contours in X,Y dimension. The general guideline for TopoToRaster is use a cell size which is half the average horizontal distance between contours. If your contours are dense enough to be ~1m apart you could use a 0.5 to 0.25 cell size and track gullies and similar features in that range.

As far what numbers to use as parameters, after choosing the cell size the defaults are pretty robust, just go with them. See creating a dem from contours for more.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.