To Maciek, Helena, Markus and all who participated in this interesting discussion on DEM,
ANUDEM, TOPOGRID and Hutchinson's Adaptive Algorithm are _not_ known to me. But I have some experience with hydraulic models on structured (raster) and unstructured (vector) grids and I would like to add some notes on DEM from a users point of view:
The quality of DEM surely is essential for any GIS analysis. Without the incooperation of general breaklines (ridges, cliffs or what ever) the DEM quality decreases. So, adding "break-line-capability" to GRASS is _very_ important.
#1
Depending on the structure of your target DEM the workflow and the tools needed will be completely different:
1. If a raster DEM is your target then tools like r.carve are useful.
2. If a vector DEM is your target a complicated procedure on combining arcs, net generation and mapping of raster onto the net is needed.
#2
From my point of view nowadays issues come from huge grids (satellite data, laserscan data, echolot data, radar data). Generally speaking there are too much data and there is the need to reduce them without loss of information. Those highres data often implicitly include general breaklines. GRASS is good in handling huge raster files. So, it would be great to extract general breaklines automatically from raster data. A first step could be to experiment with terraflow.
#3
My guess for the future is that vector DEM will stay the most efficient and popular one in engineering practice. Therefore in the long term it would be good if the bridge from raster to vector would be enlarged IN GRASS.
Karl
--------------------------------------------------------------------------
Dr.-Ing. Karl BROICH Tel: +49 / (0) 89 / 6004 2478
Universität der Bundeswehr München Fax: +49 / (0) 89 / 6004 3858
Institut für Wasserwesen
Werner-Heisenberg-Weg 39
D 85577 Neubiberg
--------------------------------------------------------------------------