[GRASS-user] Modelling for groundwater

"Sören Gebbert" soerengebbert at gmx.de
Tue Feb 6 12:38:34 EST 2007

```Hi Thomas,
i have implemented a new library (gpde) to solve
partial differential equations within grass.
Based on this library
the groundwater flow equation( based on the law of Darcy) is
discretized with the finite volume method and solved with
solvers implemented in the gpde library.

Two pde's are used, one for 2d confined and unconfined, transient
groundwater flow and one for 3d confined transient groundwater flow.

Like these ones:
http://en.wikipedia.org/wiki/Groundwater_flow_equation

The following solvers are implemented:
* method of conjugate gradients (cg) for symmetric, positive definite sparse and quadratic matrices
* method of biconjugate gradient (bicgstab) for unsymmetrical, none definite  sparse and quadratic matrices
* GAUSS elimination for quadratic matrices
* LU decomposition for quadratic matrices

All solvers and the linear equation assembling functions are parallelized
with OpenMP (www.openmp.org).

So i do not use any available groundwater flow models, i have implemented
a new one. Well, i'm quite sure that the new model is no comparison
to the existing ones (MODFLOW; FEFLOW, ...).
The current groundwater flow modules are only useful for simple problems,
only dirichlet and neumann boundary conditions are implemented.

But i hope to encourage other people to solve
pde's within grass with the gpde lib, based on raster, volume and vector data.

And a nice feature of the gpde library is the parallel computation on multicore machines. Take a look at this speedup of the cg method:
http://www-pool.math.tu-berlin.de/~soeren/grass/modules/screenshots/sgi_altix_cg_bench.png

Best regards
Soeren

-------- Original-Nachricht --------
Datum: Tue, 06 Feb 2007 12:02:53 -0500
An: "Sören Gebbert" <soerengebbert at gmx.de>
CC: Markus Neteler <neteler at itc.it>, grassuser at grass.itc.it
Betreff: Re: [GRASS-user] Modelling for groundwater

> Sören,
>
> So what model is actually used? Not MODFLOW, judging from your comments.
> I am eager to see more documentation!
>
> Regards,
> Tom
>
> Sören Gebbert wrote:
> > Hi Ravi and Markus,
> >
> >
> >> RAVI KUMAR wrote on 02/06/2007 06:34 AM:
> >>
> >>> Hi All,
> >>> Modelling for groundwater
> >>>
> >>> Pl suggest any work done through GRASS or any other FOSS GIS on the
> >>> above topic.
> >>> ThanX in anticipation
> >>> Ravi Kumar
> >>>
> >> In GRASS 6.3.cvs, recently r.gwflow and r3.gwflow were introduced:
> >> http://grass.itc.it/grass63/manuals/html63_user/r.gwflow.html
> >> http://grass.itc.it/grass63/manuals/html63_user/r3.gwflow.html
> >>
> >> Markus
> >>
> >
> > Thanks Markus,
> > right now these two modules are not documented and in a very early
> functional state, like the underlying gpde lib.
> > But i will submit in the next weeks (2-4) a major update of the gpde lib
> and the groundwater flow modules.
> > This will include a much better documentation,
> > exact null value support for 2d and 3d data arrays, array io functions,
> gradient calculation, more unit tests and
> > a new solute transport module for transport calculation of groundwater
> pollution in porous media 2d/3d.
> >
> > AFAIK, there was a module to combine modflow with grass
> > for groundwater calculation:
> > http://www.valledemexico.ambitiouslemon.com/gwmodelling.html
> > but this link is not reachable. :(
> >
> > Best regards
> > Soeren
> >
> >
> >
> >> _______________________________________________
> >> grassuser mailing list
> >> grassuser at grass.itc.it
> >> http://grass.itc.it/mailman/listinfo/grassuser
> >>
> >
> >
>
>
> --
> National Weather Service
> Ohio River Forecast Center
> 1901 South State Route 134
> Wilmington, OH 45177
>