3
<p><em>v.vol.rst</em> interpolates values to a 3-dimensional raster map from
4
3-dimensional point data (e.g. temperature, rainfall data from climatic
5
stations, concentrations from drill holes etc.) given in a 3-D vector
6
point file named <b>input</b>. The size of the output
7
3D raster map <b>elevation</b> is given by the current 3D region. Sometimes, the
9
may want to get a 2-D map showing a modelled phenomenon at a
10
crossection surface. In that case, <b>cross_input</b> and <b>cross_output</b>
11
options must be specified, with the output 2D raster map <b>cross_output</b>
12
containing the crossection of the interpolated volume with a surface
13
defined by <b>cross_input</b>
14
2D raster map. As an option, simultaneously with interpolation,
15
geometric parameters of the interpolated
16
phenomenon can be computed (magnitude of gradient, direction of
17
gradient defined by horizontal and vertical angles), change of gradient,
18
Gauss-Kronecker curvature, or mean curvature). These geometric
19
parameteres are saved as
20
3D raster maps <b>gradient, aspect_horizontal, aspect_vertical, ncurvature, gcurvature, mcurvature</b>,
21
respectively. Maps <b>aspect_horizontal</b> and <b>aspect_vertical</b> are in degrees.
23
<p>At first, data points are checked for identical positions and points
24
that are closer to each other than given <b>dmin</b> are removed.
25
Parameters <b>wmult</b> and <b>zmult</b> allow the user to re-scale
26
the w-values and z-coordinates of the point data (useful e.g. for
27
transformation of elevations given in feet to meters, so that the
28
proper values of gradient and curvatures can be computed).
29
Rescaling of z-coordinates (<b>zmult</b>) is also needed when the distances
30
in vertical direction are much smaller than the horizontal
31
distances; if that is the case, the value of <b>zmult</b>
32
should be selected so that the vertical and horizontal distances
33
have about the same magnitude.
35
<p>Regularized spline with tension method is used in the interpolation.
36
The <b>tension</b> parameter controls the distance over which
37
each given point influences the resulting volume (with very high tension,
38
each point influences only its close neighborhood and the volume goes
39
rapidly to trend between the points).
40
Higher values of tension parameter reduce the overshoots that
41
can appear in volumes with rapid change of gradient. For noisy data, it
42
is possible to define a global smoothing parameter, <b>smooth</b>.
44
smoothing parameter set to zero (<b>smooth=0</b>) the resulting volume
45
passes exactly through the data points.
46
When smoothing is used, it is possible to output a vector map <b>deviations</b>
47
containing deviations of the resulting volume from the given data.
48
<p>The user can define a 2D raster map named <b>maskmap</b>, which will
49
be used as a mask. The interpolation is skipped for 3-dimensional cells
50
whose 2-dimensional projection has a zero value in the mask. Zero values will
51
be assigned to these cells in all output 3D raster maps.
52
<p>If the number of given points is greater than 700, segmented
53
processing is used. The region is split into 3-dimensional "box"
54
segments, each having less than <b>segmax</b> points and interpolation
55
is performed on each segment of the region. To ensure the smooth
56
connection of segments, the interpolation function for each segment is
57
computed using the points in the given segment
58
and the points in its neighborhood. The minimum number of points taken
59
for interpolation is controlled by <b>npmin</b> , the value of which
60
must be larger than <b>segmax</b> and less than 700. This limit of 700 was
61
selected to ensure the numerical stability and efficiency of the
67
Using the <b>where</b> parameter, the interpolation can be limited to use
68
only a subset of the input vectors.
70
<div class="code"><pre>
71
# preparation as in above example
72
v.vol.rst elevrand_3d wcol=soilrange elevation=soilrange zmult=100 where="soilrange > 3"
76
<h3>Cross validation procedure</h3>
78
Sometimes it can be difficult to figure out the proper values of
79
interpolation parameters. In this case, the user can use a
80
crossvalidation procedure using <b>-c</b> flag (a.k.a. "jack-knife"
81
method) to find optimal parameters for given data. In this method,
82
every point in the input point file is temporarily excluded from the
83
computation and interpolation error for this point location is
84
computed. During this procedure no output grid files can be
85
simultanuously computed. The procedure for larger datasets may take a
86
very long time, so it might be worth to use just a sample data
87
representing the whole dataset.
90
(based on <a href="http://www.grassbook.org/data_menu2nd.php">Slovakia3d dataset</a>):</i>
91
<p><div class="code"><pre>
93
g.region n=5530000 s=5275000 w=4186000 e=4631000 res=500 -p
94
v.vol.rst -c input=precip3d wcolumn=precip zmult=50 segmax=700 cvdev=cvdevmap tension=10
96
v.univar cvdevmap col=flt1 type=point
99
Based on these results, the parameters will have to be optimized. It is
100
recommended to plot the CV error as curve while modifying
102
<p>The best approach is to start with <b>tension</b>, <b>smooth</b>
103
and <b>zmult</b> with rough steps, or to set <b>zmult</b> to a
104
constant somewhere between 30-60. This helps to find minimal RMSE
105
values while then finer steps can be used in all parameters. The
106
reasonable range is <b>tension</b>=10...100,
107
<b>smooth</b>=0.1...1.0, <b>zmult</b>=10...100.
108
<p>In <em>v.vol.rst</em> the tension parameter is much more sensitive to
109
changes than in <em>v.surf.rst</em>,
110
therefore the user should always check the
111
result by visual inspection. Minimizing CV does not always provide the best
112
result, especially when the density of data are insufficient. Then
113
the optimal result found by CV is an oversmoothed surface.
116
The vector points map must be a 3D vector map (x, y, z as geometry).
117
The module <a href="v.in.db.html">v.in.db</a> can be used to generate
118
a 3D vector map from a table containing x,y,z columns.
120
Also, the input data should be in a projected coodinate system, such as
121
Univeral Transverse Mercator. The module does not appear to have support for
122
geographic (Lat/Long) coordinates as of May 2009.
124
<p><em>v.vol.rst</em> uses regularized spline with tension for
125
interpolation from point data (as described in Mitasova and Mitas,
126
1993). The implementation has an improved segmentation procedure based
127
on Oct-trees which enhances the efficiency for large data sets.
129
<p>Geometric parameters - magnitude of gradient (<b>gradient</b>),
130
horizontal (<b>aspect_horizontal</b>) and vertical (<b>aspect_vertical)</b>aspects,
131
change of gradient (<b>ncurvature</b>), Gauss-Kronecker (<b>gcurvature</b>) and
132
mean curvatures (<b>mcurvature</b>) are computed directly from the
133
interpolation function so that the important relationships between
134
these parameters are preserved. More information on these parameters
135
can be found in Mitasova et al., 1995 or Thorpe, 1979.
137
<p>The program gives warning when significant overshoots appear and
138
higher tension should be used. However, with tension too high the
139
resulting volume will have local maximum in each given point
140
and everywhere else the volume goes rapidly to trend. With a smoothing
141
parameter greater than zero, the volume will not pass through the data
142
points and the higher the parameter the closer the volume will be to the
143
trend. For theory on smoothing with splines see Talmi and Gilat, 1977 or Wahba, 1990.
145
<p>If a visible connection of segments appears, the program should be
146
rerun with higher <b>npmin</b> to get more points from the
147
neighborhood of given segment.
149
<p>If the number of points in a vector map is less than 400, <b>segmax</b>
150
should be set to 400 so that segmentation is not performed when it is
153
<p>The program gives a warning when the user wants to interpolate outside the
154
"box" given by minimum and maximum coordinates in the input vector map.
155
To remedy this, zoom into the area encompassing the input vector data points.
157
<p>For large data sets (thousands of data points), it is suggested to
158
zoom into a smaller representative area and test whether the parameters
159
chosen (e.g. defaults) are appropriate.
161
<p>The user must run <em>g.region</em> before the program to set the
162
3D region for interpolation.
167
<!-- TODO: find better data. This example is nonsensical :-) -->
168
Spearfish example (we first simulate 3D soil range data):
170
<div class="code"><pre>
173
g.region res=100 tbres=100 res3=100 b=0 t=1500 -ap3
175
### First part: generate synthetic 3D data (true 3D soil data preferred)
176
# generate random positions from elevation map (2D)
177
r.random elevation.10m vector_output=elevrand n=200
179
# generate synthetic values
180
v.db.addcolumn elevrand col="x double precision, y double precision"
181
v.to.db elevrand option=coor col=x,y
185
v.in.db elevrand out=elevrand_3d x=x y=y z=value key=cat
186
v.info -c elevrand_3d
187
v.info -t elevrand_3d
189
# remove the now superfluous 'x', 'y' and 'value' (z) columns
190
v.db.dropcolumn elevrand_3d col=x
191
v.db.dropcolumn elevrand_3d col=y
192
v.db.dropcolumn elevrand_3d col=value
194
# add attribute to have data available for 3D interpolation
195
# (Soil range types taken from the USDA Soil Survey)
199
v.db.addcolumn elevrand_3d col="soilrange integer"
200
v.what.rast elevrand_3d col=soilrange rast=soils.range
202
# fix 0 (no data in raster map) to NULL:
203
v.db.update elevrand_3d col=soilrange value=NULL where="soilrange=0"
204
v.db.select elevrand_3d
206
# optionally: check 3D points in Paraview
207
v.out.vtk input=elevrand_3d output=elevrand_3d.vtk type=point dp=2
208
paraview --data=elevrand_3d.vtk
210
### Second part: 3D interpolation from 3D point data
211
# interpolate volume to "soilrange" voxel map
212
v.vol.rst input=elevrand_3d wcol=soilrange elevation=soilrange zmult=100
214
# visualize I: in GRASS GIS wxGUI
216
# load: 2D raster map: elevation.10m
217
# 3D raster map: soilrange
219
# visualize II: export to Paraview
220
r.mapcalc "bottom = 0.0"
221
r3.out.vtk -s input=soilrange top=elevation.10m bottom=bottom dp=2 output=volume.vtk
222
paraview --data=volume.vtk
226
<b>deviations</b> file is written as 2D and deviations are not written as attributes.
229
<p>Hofierka J., Parajka J., Mitasova H., Mitas L., 2002, Multivariate
230
Interpolation of Precipitation Using Regularized Spline with Tension.
232
GIS 6, pp. 135-150.
233
<p><a href="http://www4.ncsu.edu/~hmitaso/gmslab/">Mitas, L.,
234
Mitasova, H.</a>, 1999, Spatial Interpolation. In: P.Longley, M.F.
235
Goodchild, D.J. Maguire, D.W.Rhind (Eds.), Geographical Information
236
Systems: Principles, Techniques, Management and Applications, Wiley,
238
<p>Mitas L., Brown W. M., Mitasova H., 1997,
239
<a href="http://www4.ncsu.edu/~hmitaso/gmslab/lcgfin/cg-mitas.html">Role
240
of dynamic cartography in simulations of landscape processes based on
241
multi-variate fields.</a> Computers and Geosciences, Vol. 23, No. 4,
242
pp. 437-446 (includes CDROM and WWW: www.elsevier.nl/locate/cgvis)
243
<p>Mitasova H., Mitas L., Brown W.M., D.P. Gerdes, I.
244
Kosinovsky, Baker, T.1995, Modeling spatially and temporally
245
distributed phenomena:
246
New methods and tools for GRASS GIS. International Journal of GIS, 9
248
special issue on Integrating GIS and Environmental modeling, 433-446.
249
<p> Mitasova, H., Mitas, L., Brown, B., Kosinovsky, I., Baker, T.,
251
<a href="http://www4.ncsu.edu/~hmitaso/gmslab/viz/ches.html">Multidimensional
252
interpolation and visualization in GRASS GIS</a>
253
<p><a href="http://www4.ncsu.edu/~hmitaso/gmslab/papers/lmg.rev1.ps">Mitasova
254
H. and Mitas L. 1993</a>: Interpolation by Regularized Spline with
255
Tension: I. Theory and Implementation, <i>Mathematical Geology</i> 25,
257
<p><a href="http://www4.ncsu.edu/~hmitaso/gmslab/papers/hmg.rev1.ps">Mitasova
258
H. and Hofierka J. 1993</a>: Interpolation by Regularized Spline with
259
Tension: II. Application to Terrain Modeling and Surface Geometry
260
Analysis, <i>Mathematical Geology</i> 25, 657-667.
261
<p>Mitasova, H., 1992 : New capabilities for interpolation and
262
topographic analysis in GRASS, GRASSclippings 6, No.2 (summer), p.13.
263
<p>Wahba, G., 1990 : Spline Models for Observational Data, CNMS-NSF
264
Regional Conference series in applied mathematics, 59, SIAM,
265
Philadelphia, Pennsylvania.
266
<p>Mitas, L., Mitasova H., 1988 : General variational approach to the
267
interpolation problem, Computers and Mathematics with Applications 16,
269
<p>Talmi, A. and Gilat, G., 1977 : Method for Smooth Approximation of
270
Data, Journal of Computational Physics, 23, p.93-123.
271
<p>Thorpe, J. A. (1979): Elementary Topics in Differential Geometry.
272
Springer-Verlag, New York, pp. 6-94.
277
<a href="g.region.html">g.region</a>,
278
<a href="v.in.ascii.html">v.in.ascii</a>,
279
<a href="r3.mask.html">r3.mask</a>,
280
<a href="v.in.db.html">v.in.db</a>,
281
<a href="v.surf.rst.html">v.surf.rst</a>,
282
<a href="v.univar.html">v.univar</a>
287
Original version of program (in FORTRAN) and GRASS enhancements: <br>
288
Lubos Mitas, NCSA, University of Illinois at Urbana-Champaign,
289
Illinois, USA, since 2000 at Department of Physics,
290
North Carolina State University, Raleigh, USA
291
<a href="mailto:lubos_mitas@ncsu.edu">lubos_mitas@ncsu.edu</a>
293
Helena Mitasova, Department of Marine, Earth and Atmospheric Sciences,
294
North Carolina State University, Raleigh, USA,
295
<a href="mailto:hmitaso@unity.ncsu.edu">hmitaso@unity.ncsu.edu</a>
297
Modified program (translated to C, adapted for GRASS, new
298
segmentation procedure): <br>
299
Irina Kosinovsky, US Army CERL, Champaign, Illinois, USA <br>
300
Dave Gerdes, US Army CERL, Champaign, Illinois, USA
302
Modifications for g3d library, geometric parameters,
303
cross-validation, deviations: <br>
304
Jaro Hofierka, Department of Geography and Regional Development,
305
University of Presov, Presov, Slovakia,
306
<a href="mailto:hofierka@fhpv.unipo.sk">hofierka@fhpv.unipo.sk</a>,
307
<a href="http://www.geomodel.sk">http://www.geomodel.sk</a>
309
<p><i>Last changed: $Date: 2014-11-28 17:25:40 +0100 (Fri, 28 Nov 2014) $</i>