[GRASS-user] r.watershed and basins that are sinks

[M S]

I've had good results with r.watershed, however, I'm seeking further
clarification about the 'depression' input, specifically when the watershed
of interest is an internally drained basin.

I have a watershed that is an internally drained basin, where most of the
surface water outflows as groundwater.  The basin is in the order of about
13 square miles.  If I don't input a depression dataset, I get a nice
delineation of the sink basin, with (what appears to be only) one
exception:  At the lowest point along the ridge of the internally drained
basin, r.watershed wants to bump out the divide delineation to another low
point in the DEM.  In other words, inside the large sink basin, I may have
elevation as low as 15 feet.  The highest point before overflowing the basin
is around 31 feet.  At this pop-off point for the sink, it seems logical
that a divide should be there.  Instead, it does not put a divide there, and
swings the delineation out to encompass another low spot (about 15 feet
elevation) on the other side of the 31 foot divide.

It seems that, even with an internally drained basin, it wants to find and
incorporate the lowest point to drain to, regardless of the fact that it
needs filled about 15 feet before it would flow to that side of the divide.

If I input data from the r.lake module as "depression", I only get output
maps from r.watershed that are all populated with zeros (black), although I
am still trying various inputs for the depression map.   I am not sure that
this input parameter is applicable for defining an internally drained basin,
as it is to define depressions that may retain water during a rain event.
Even though the basin can be filled up to about 31 feet with r.lake before
it pops-off, there are still dry spots in the internal basin that do not get
totally flooded, meaning they are higher than 31 foot.  Another trial will
be to make sure the depression map that is inputted has no "upland islands"
in it, but is all contiguous area for the depression.

My questions are:
Is it appropriate to use something like r.lake to fill the basin to the
pop-off level, and then use that as input into r.watershed via the
"depression" parameter?
Are there any other special considerations I should be aware of to model
basins that are actually real sinks?
Are there other GRASS tools I should be looking at to aid in this internal
basin delineation?



Thanks,
Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/grass-user/attachments/20080520/2b566658/attachment.html