Note: A new GRASS GIS stable version has been released: GRASS GIS 7, available here. And see the new manual page here
r.sim.water - Overland flow hydrologic simulation using path sampling method (SIMWE).
raster, flow, hydrology
r.sim.water [-t] elevin=name dxin=name dyin=name [rain=name] [rain_val=float] [infil=name] [infil_val=float] [manin=name] [manin_val=float] [traps=name] [depth=name] [disch=name] [err=name] [nwalk=integer] [niter=integer] [outiter=integer] [diffc=float] [hmax=float] [halpha=float] [hbeta=float] [--overwrite] [--verbose] [--quiet]
- Time-series output
- Allow output files to overwrite existing files
- Verbose module output
- Quiet module output
- Name of elevation raster map [m]
- Name of x-derivatives raster map [m/m]
- Name of y-derivatives raster map [m/m]
- Name of rainfall excess rate (rain-infilt) raster map [mm/hr]
- Rainfall excess rate unique value [mm/hr]
- Default: 50
- Name of runoff infiltration rate raster map [mm/hr]
- Runoff infiltration rate unique value [mm/hr]
- Default: 0.0
- Name of Manning's n raster map
- Manning's n unique value
- Default: 0.1
- Name of flow controls raster map (permeability ratio 0-1)
- Name for output water depth raster map [m]
- Name for output water discharge raster map [m3/s]
- Name for output simulation error raster map [m]
- Number of walkers, default is twice the number of cells
- Time used for iterations [minutes]
- Default: 10
- Time interval for creating output maps [minutes]
- Default: 2
- Water diffusion constant
- Default: 0.8
- Threshold water depth [m]
- Diffusion increases after this water depth is reached
- Default: 0.3
- Diffusion increase constant
- Default: 4.0
- Weighting factor for water flow velocity vector
- Default: 0.5
r.sim.water is a landscape scale simulation model
of overland flow designed for spatially variable terrain, soil, cover
and rainfall excess conditions. A 2D shallow water flow is described by
the bivariate form of Saint Venant equations. The numerical solution is based
on the concept of duality between the field and particle representation of
the modeled quantity. Green's function Monte Carlo method, used to solve the equation,
provides robustness necessary for spatially variable conditions and high
resolutions (Mitas and Mitasova 1998). The key inputs of the model include
elevation (elevin raster map), flow gradient vector given by
first-order partial derivatives of elevation field (dxin and dyin
raster maps), rainfall excess rate (rain raster map or rain_val single
value) and a surface roughness coefficient given by Manning's n
(manin raster map or manin_val single value). Partial
derivatives raster maps can be computed along with interpolation of a DEM using
the -d option in v.surf.rst module. If elevation raster
map is already provided, partial derivatives can be computed using
r.slope.aspect module. Partial derivatives are used
to determine the direction and magnitude of water flow velocity. To include a
predefined direction of flow, map algebra can be used to replace terrain-derived
partial derivatives with pre-defined partial derivatives in selected grid cells such
as man-made channels, ditches or culverts. Equations (2) and (3) from
can be used to compute partial derivates of the predefined flow using its direction given
by aspect and slope.
The module automatically converts horizontal distances from feet to metric system using
database/projection information. Rainfall excess is defined as rainfall intensity
- infiltration rate and should be provided in [mm/hr].
Rainfall intensities are usually available from meteorological stations.
Infiltration rate depends on soil properties and land cover. It varies in space and time.
For saturated soil and steady-state water flow it can be estimated using
saturated hydraulic conductivity rates based on field measurements or using
reference values which can be found in literature.
Optionally, user can provide an overland flow infiltration rate map
infil or a single value infil_val in [mm/hr] that control the rate of
infiltration for the already flowing water, effectively reducing the flow depth and
Overland flow can be further controled by permeable check dams or similar type of structures,
the user can provide a map of these structures and their permeability ratio
in the map traps that defines the probability of particles to pass
through the structure (the values will be 0-1).
Output includes a water depth raster map depth in [m], and a water discharge
raster map disch in [m3/s]. Error of the numerical solution can be analyzed using
the err raster map (the resulting water depth is an average, and err is its RMSE).
The output vector points map outwalk can be used to analyze and visualize
spatial distribution of walkers at different simulation times (note that
the resulting water depth is based on the density of these walkers). Number
of the output walkers is controled by the density parameter, which controls
how many walkers used in simulation should be written into the output.
Duration of simulation is controled by the niter parameter. The default value
is 10 minutes, reaching the steady-state may require much longer time,
depending on the time step, complexity of terrain, land cover and size of the area.
Output water depth and discharge maps can be saved during simulation using
the time series flag -t and outiter parameter
defining the time step in minutes for writing output files.
Files are saved with a suffix representing time since the start of simulation in seconds
(e.g. wdepth.500, wdepth.1000).
Overland flow is routed based on partial derivatives of elevation
field or other landscape features influencing water flow. Simulation
equations include a diffusion term (diffc parameter) which enables
water flow to overcome elevation depressions or obstacles when water depth exceeds
a threshold water depth value (hmax), given in [m]. When it is reached,
diffusion term increases as given by halpha and advection term
(direction of flow) is given as "prevailing" direction of flow computed
as average of flow directions from the previous hbeta number of grid cells.
A 2D shallow water flow is described by the bivariate form of Saint
Venant equations (e.g., Julien et al., 1995). The continuity of water
flow relation is coupled with the momentum conservation equation and
for a shallow water overland flow, the hydraulic radius is approximated
by the normal flow depth. The system of equations is closed using the
Manning's relation. Model assumes that the flow is close to the kinematic
wave approximation, but we include a diffusion-like term to incorporate the
impact of diffusive wave effects. Such an incorporation of diffusion
in the water flow simulation is not new and a similar term has been obtained
in derivations of diffusion-advection equations for overland flow, e.g.,
by Lettenmeier and Wood, (1992). In our reformulation, we simplify the
diffusion coefficient to a constant and we use a modified diffusion term.
The diffusion constant which we have used is rather small (approximately
one order of magnitude smaller than the reciprocal Manning's coefficient)
and therefore the resulting flow is close to the kinematic regime. However,
the diffusion term improves the kinematic solution, by overcoming small
shallow pits common in digital elevation models (DEM) and by smoothing out
the flow over slope discontinuities or abrupt changes in Manning's coefficient
(e.g., due to a road, or other anthropogenic changes in elevations or cover).
Green's function stochastic method of solution.
The Saint Venant equations are solved by a stochastic method called Monte Carlo
(very similar to Monte Carlo methods in computational fluid dynamics or to
quantum Monte Carlo approaches for solving the Schrodinger equation (Schmidt
and Ceperley, 1992, Hammond et al., 1994; Mitas, 1996)). It is assumed
that these equations are a representation of stochastic processes with
diffusion and drift components (Fokker-Planck equations).
The Monte Carlo technique has several unique advantages which are
becoming even more important due to new developments in computer technology.
Perhaps one of the most significant Monte Carlo properties is robustness
which enables us to solve the equations for complex cases, such as discontinuities
in the coefficients of differential operators (in our case, abrupt slope
or cover changes, etc). Also, rough solutions can be estimated rather
quickly, which allows us to carry out preliminary quantitative studies
or to rapidly extract qualitative trends by parameter scans. In addition,
the stochastic methods are tailored to the new generation of computers
as they provide scalability from a single workstation to large parallel
machines due to the independence of sampling points. Therefore, the methods
are useful both for everyday exploratory work using a desktop computer and
for large, cutting-edge applications using high performance computing.
g.region rast=elevation.10m -p
r.slope.aspect elevation=elevation.10m dx=elev_dx dy=elev_dy
# synthetic maps
r.mapcalc "rain = if(elevation.10m, 5.0, null())"
r.mapcalc "manning = if(elevation.10m, 0.05, null())"
r.mapcalc "infilt = if(elevation.10m, 0.0, null())"
r.sim.water elevin=elevation.10m dxin=elev_dx dyin=elev_dy \
rain=rain manin=manning infil=infilt \
d.rast.leg depth pos=85
d.his i=elevation.10m.shade h=depth
d.barscale at=4,92 bcolor=none tcolor=black -t
Water depth map in the Spearfish (SD) area
If the module fails with
then a lower nwalk parameter value has to be selected.
ERROR: nwalk (7000001) > maxw (7000000)!
Helena Mitasova, Lubos Mitas
North Carolina State University
GeoModel, s.r.o. Bratislava, Slovakia
North Carolina State University
- Mitasova, H., Thaxton, C., Hofierka, J., McLaughlin, R., Moore, A., Mitas L., 2004,
Path sampling method for modeling overland water flow, sediment transport
and short term terrain evolution in Open Source GIS.
In: C.T. Miller, M.W. Farthing, V.G. Gray, G.F. Pinder eds.,
Proceedings of the XVth International Conference on Computational Methods in Water
Resources (CMWR XV), June 13-17 2004, Chapel Hill, NC, USA, Elsevier, pp. 1479-1490.
- Mitasova H, Mitas, L., 2000,
processes in multiscale framework: exploring duality between particles and fields,
plenary talk at GIScience2000 conference, Savannah, GA.
- Mitas, L., and Mitasova, H., 1998, Distributed soil erosion simulation
for effective erosion prevention. Water Resources Research, 34(3), 505-516.
- Mitasova, H., Mitas, L., 2001,
Multiscale soil erosion simulations for land use management,
In: Landscape erosion and landscape evolution modeling, Harmon R. and Doe W. eds.,
Kluwer Academic/Plenum Publishers, pp. 321-347.
- Neteler, M. and Mitasova, H., 2008,
Open Source GIS: A GRASS GIS Approach. Third Edition.
The International Series in Engineering and Computer Science: Volume 773. Springer New York Inc, p. 406.
Last changed: $Date: 2014-03-15 07:28:47 -0700 (Sat, 15 Mar 2014) $
Main index - raster index - Full index
© 2003-2016 GRASS Development Team