TransDEM Forum

G'day geophil,

Roland, I believe I have come across the situation where the latest Aster GDEM data that I have downloaded for a route upon which I am working has 'failed miserably' in producing a DEM in TransDEM with the proper heights. I understand that I can use the process espoused in Tutorial 10 to overcome this (you're a very clever man, you know that, don't you?) but there are a few aspects of this process that are 'unclear' to me, so I thought I would go to the most likely source to remove this 'doubt', before I 'jump in at the deep end'...

...firstly, is it essential to use five Control Points, as indicated in the tutorial or can the process be accomplished with less (obviously, more, would be, I guess, a 'bonus')?...

...and do these Control Points have to be as close to the edges of the route's DEM as possible (or will the entire route be 'adjusted', accordingly, even if the Control Points are within a small area of the whole 'space' occupied by the entire DEM)?...

Jerker {:)}

Jerker wrote:...firstly, is it essential to use five Control Points, as indicated in the tutorial or can the process be accomplished with less (obviously, more, would be, I guess, a 'bonus')?...

Garry,

for the triangular network to work as originally intended you will need three or more control points. Pages 20 to 22 in the TransDEM main manual have a bit more information: "Control Points and Triangular Networks".

The original intention was to adjust elevations with a mesh of triangles. However, the concept can also be used be used differently, like adjusting an entire DEM as in tutorial 10. With a single control point, you can raise or lower the DEM as a whole, with two points you create something like an inclined plane for adjustment. Three control points are the minimum for the triangular network: the single triangle. With five points you will have three or four triangles. (TransDEM will show their edges.)

This tutorial 10 was written for the first version of ASTER DEMs which often showed severe errors, often due to vertical miscalibration. You say your current DEM is an ASTER GDEM (which would be "version 2"). As far as I know these were constructed by merging the original ("version 1") granules, but with vertical correction applied. This means the overall or average elevation level should be more or less correct. Artefacts, however, are still quite common in GDEM, some with a significant extent.

Jerker wrote:...and do these Control Points have to be as close to the edges of the route's DEM as possible (or will the entire route be 'adjusted', accordingly, even if the Control Points are within a small area of the whole 'space' occupied by the entire DEM)?...

The more sophisticated "barycentric" interpolation algorithm works best within a "bounded region", spanned by the outer control points, if all in "fix" mode. Outside this bounded region (the "convex hull") no re-processing will occur. You can and should set control point mode accordingly. At the edges of the DEM, auxiliary control points will always be created automatically, but this outside area is less controllable. Some gaps at or very near the edges may exist.

In addition to the control point / triangular network feature, TransDEM offers the "Raise/Lower DEM" function (page 70) which transforms an input elevation range to an output elevation range. This function was also inspired by early ASTER DEMs.