1. I have had Ph.D.’s in mathematics argue both sides of this issue.
  2. GIS pros are more comfortable with a raster DTM, while engineers prefer the TIN method.
  3. Choose the method that best suits the problem and preserves data integrity.

As a disclaimer, my intent in this post is to stimulate thought and discussion, not to arrive at a definitive conclusion. I have had Ph.D.’s in mathematics take both sides of this debate, so far be it from me to claim I know the right answer. Also, I am assuming that the reader knows the difference between the two, but I will provide some general references, just in case.

From the practical side in most of today’s popular CAD and GIS software there is a relatively small – by comparison to point clouds, limit to the size of a file that can be processed. In AutoCAD for instance that limit is generally between  2 to 5 million points, without some form of external help. Obviously if you need a continuous surface model, let’s say for a large drainage study of 100’s of millions of points, then DTM is probably your only choice.

On the other hand, engineers in general prefer to use the TIN method. They want to know that the elevation at each break point was held and used in the solution. They want control.

This is where I think the conflict can begin to arise. An aerial LiDAR survey “samples” the terrain in a way that is much more like a grid than an irregular network. Remember scanners are dumb. It seems to me that the TIN method was designed to work with a collection of points randomly located in the field by a survey crew. That makes TIN and manual survey a good fit spatially.

The real problems begin to arise when the user begins to derive products from the original survey data, regardless of how it was obtained. There is always error in every measurement. It is when the user creates a TIN from a DTM, so that they can create contours, that one begins to question the validity of the workflow.

One of the potential solutions is the use of breaklines. In a drainage application, such as a highway widening project, creation of breaklines from the LiDAR data set could provide the TIN with the input needed to derive a more appropriate solution than just using the raw LiDAR data. I think this would tend to preserve the integrity of the original data.

In general, the GIS world is more comfortable with a raster DTM, usually because it can be more intelligently rendered  for visualization purposes, such as hill shading. The survey/engineering world prefers a TIN as the basis for their drainage and volume computations, but I have been told that in some cases the opposite approach will produce more accurate results. It would seem that a stockpile quantity survey would be more accurately performed using a DTM of tightly spaced points.

As is so often the case the answer is, it depends. I think the key is to select the method that is most appropriate to the problem you are trying to solve, while keeping an eye on error propagation and data integrity.

What do you think?

This entry was posted in Technology and tagged , . Bookmark the permalink.

3 Responses to DTM vs. TIN

  1. Ryan says:

    I think you’ve hit the nail on the head with “it depends”! In addition to the very valid points you’ve raised, there are a couple of other considerations as well:

    – Redundancy: the beauty of TINs is that you can model relatively flat areas with few points and rugged terrain with as much density/accuracy as you want. This has advantages over regular-density grids (file size, etc).
    – Accuracy: if you create a 5 meter grid and then decide you really need 2 meter post spacing to achieve accuracy requirements, you’re in trouble. With TINs you can maintain existing breaklines and throw in more points/breaklines to achieve accuracy requirements.
    – Format: raster formats tend to be commonly understood between software applications, whereas there is no common TIN format. Thus, TINs aren’t as transportable as grids, and because the formats are proprietary one needs to question whether or not it would be better to archive the data as a grid (or ASCII with point codes for breaklines).
    – Workflow: some operators may be more comfortable with TIN versus grid editing.

  2. Gene V. Roe says:

    Thanks Ryan – excellent input.

  3. Zana says:

    Great article! Thank you

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.