[GRASS-dev] Need ideas for a masters thesis...

vetter michael mv at ipf.tuwien.ac.at
Mon Nov 16 05:07:53 EST 2009

We already did  a method to derive river centerlines based on r.thin and 
a method to calculate cross sections (see the scripts). It is not a 
method to derive the flow line of the river, but you can calculate the 
center line of the river.

For those you need a very precise water surface polygon. To derive such 
a high quality water surface you need a ortho image or a good DTM like a 
1m LiDAR DTM. By doing a classification using region growing of height 
differences and the slope.

you can check out the scripts here:

The sand or gravel banks are really a big problem in deriving a water 
surface. Such a solution is presented in:
Höfle, B., Vetter**, M., Pfeifer, N., Mandlburger, G., Stötter, J. 
*(2009)*: Water surface mapping from airborne laser scanning using 
signal amplitude and elevation data. Earth Surface Processes and 
Landforms, p. 1635-1649. DOI:10.1002/esp.1853.
good luck,

> Thomas wrote:
>> If you come from a hydraulics background… one very
>> important need is the ability to generate channel
>> cross-sections interactively for, say, HEC-RAS and then do
>> flood inundation mapping from HEC-RAS from the modeling
>> results — I would think this would be a huge
>> contribution.
> to get a good cross section you need to set it perpendicular to the 
> river's
> center line. Which means you need a good river centerline. Which means 
> you
> have to deal with the "river mile problem".
> raster tools are not always good here as the river can be very thin
> compared to the overall length leading to untenable resolution settings
> to get an good result. (r.cost + r.param.scale to pull out the cost
> ridges or r.slope.aspect to pull out the maxima from the 1st derivative
> of the cost works, but the resolution issue gets ya)
> any thoughts? some sort of reverse v.buffer?
> small islands within the river also cause havoc with many computational
> approaches.
> Hamish

More information about the grass-dev mailing list