2
<em>v.surf.bspline</em> performs a bilinear/bicubic spline interpolation with
3
Tykhonov regularization. The input is a 2D or 3D vector points map. Values to
4
interpolate can be the z values of 3D points or the values in a user-specified
5
attribue column in a 2D or 3D map. Output can be a raster or vector map.
6
Optionally, a "sparse point" vector map can be input which indicates the
7
location of <b><i>output</i></b> vector points.
9
From a theoretical perspective, the interpolating procedure takes place in two
10
parts: the first is an estimate of the linear coefficients of a spline function
11
is derived from the observation points using a least squares regression; the
12
second is the computation of the interpolated surface (or interpolated vector
13
points). As used here, the splines are 2D piece-wise non-zero polynomial
14
functions calculated within a limited, 2D area. The length of each spline step
15
is defined by <b><i>sie</i></b> for the east-west direction and
16
<b><i>sin</i></b> for the north-south direction. For optimum performance, the
17
length of spline step should be no less than the distance between observation
18
points. Each vector point observation is modeled as a linear function of the
19
non-zero splines in the area around the observation. The least squares
20
regression predicts the the coefficients of these linear functions.
21
Regularization, avoids the need to have one one observation and one coefficient
22
for each spline (in order to avoid instability).
25
With regularly distributed data points, a spline step corresponding to the
26
maximum distance between two points in both the east and north directions is
27
sufficient. But often data points are not regularly distributed and require
28
statistial regularization or estimation. In such cases, v.surf.bspline will
29
attempt to minimize the gradient of bilinear splines or the curvature of bicubic
30
splines in areas lacking point observations. As a general rule, spline step
31
length should be greater than the mean distance between observation points
32
(twice the distance between points is a good starting point). Separate east-west
33
and north-south spline step length arguments allows the user to account for some
34
degree of anisotropy in the distribution of observation points. Short spline
35
step lengths--especially spline step lengths that are less than the distance
36
between observation points--can greatly increase processing time.
39
Moreover, the maximum number of splines for each direction at each time is
40
fixed, regardless of the spline step length. As the total number of splines used
41
increases (i.e., with small spline step lengths), the region is automatically
42
into subregions for interpolation. Each subregion can contain no more than
43
150x150 splines. To avoid subregion boundary problems, subregions are created to
44
partially overlap each other. A weighted mean of observations, based on point
45
locations, is calculated within each subregion.
48
The Tykhonov regularization parameter ("<b><i>lambda_i</i></b>") acts to smooth
49
the interpolation. With a small <b><i>lambda_i</i></b>, the interpolated surface
50
closely follows observation points; a larger value will produce a smoother
54
The input can be a 2D pr 3D vector points map. If "<b><i>layer =</i></b>" 0 the
55
z-value of a 3D map is used for interpolation. If layer > 0, the user must
56
specify an attribute column to used for interpolation using the
57
"<b><i>column=</i></b>" argument (2D or 3D map).
60
v.surf.bspline can produce a raster OR a vector output (NOT simultaneously).
61
However, a vector output cannot be obtained using the default GRASS DBF driver.
64
If output is a vector points map and a "<b><i>sparse=</i></b>" vector points map
65
is not specified, the output vector map will contain points at the same
66
locations as observation points in the input map, but the values of the output
67
points are interpolated values. If instead a "<b><i>sparse=</i></b>" vector points
68
map is specified, the output vector map will contain points at the same locations as
69
the sparse vector map points, and values will be those of the interpolated
70
raster surface at those points.
73
A cross validation "leave-one-out" analysis is available to help to determine
74
the optimal <b><i>lambda_i</i></b> value that produces an interpolation that
75
best fits the original observation data. The more points used for
76
cross-validation, the longer the time needed for computation. Empirical testing
77
indicates a threshold of a maximum of 100 points is recommended. Note that cross
78
validation can run very slowly if more than 100 observations are used. The
79
cross-validation output reports <i>mean</i> and <i>rms</i> of the residuals from
80
the true point value and the estimated from the interpolation for a fixed series
81
of <b><i>lambda_i</i></b> values. No vector nor raster output will be created
82
when cross-validation is selected.
86
<h3>Basic interpolation</h3>
88
<div class="code"><pre>
89
v.surf.bspline input=point_vector output=interpolate_surface method=bicubic
92
A bicubic spline interpolation will be done and a vector points map with estimated
93
(i.e., interpolated) values will be created.
95
<h4>Basic interpolation and raster output with a longer spline step</h4>
97
<div class="code"><pre>
98
v.surf.bspline input=point_vector raster=interpolate_surface sie=25 sin=25
101
A bilinear spline interpolation will be done with a spline step length of 25 map
102
units. An interpolated raster map will be created at the current region resolution.
104
<h4>Estimation of <b><i>lambda_i</i></b> parameter with a cross validation proccess</h4>
106
<div class="code"><pre>
107
v.surf.bspline -c input=point_vector
110
<h4>Estimation on sparse points</h4>
112
<div class="code"><pre>
113
v.surf.bspline input=point_vector sparse=sparse_points output=interpolate_surface
116
An output map of vector points will be created, corresponding to the sparse vector map, with interpolated values.
118
<h4>Using attribute values instead Z-coordinates</h4>
119
<div class="code"><pre>
120
v.surf.bspline input=point_vector raster=interpolate_surface layer=1 column=attrib_column
123
The interpolation will be done using the values in attrib_column, in the
124
table associated with layer 1.
129
In order to avoid RAM memory problems, an auxiliary table is needed for
130
recording some intermediate calculations. This requires the "<b>GROUP BY</b>"
131
SQL function is used, which is not supported by the "<b>dbf</b>" driver. For
132
this reason, vector map output "<b><i>output=</i></b>" is not permitted with the
133
DBF driver. There are no problems with the raster map output from the DBF
139
<a href="v.surf.idw.html">v.surf.idw</a>,
140
<a href="v.surf.rst.html">v.surf.rst</a>
145
Original version in GRASS 5.4: (s.bspline.reg)
147
Maria Antonia Brovelli, Massimiliano Cannata, Ulisse Longoni, Mirko Reguzzoni
149
Update for GRASS 6.X and improvements:
155
Brovelli M. A., Cannata M., and Longoni U.M., 2004, LIDAR Data
156
Filtering and DTM Interpolation Within GRASS, Transactions in GIS,
157
April 2004, vol. 8, iss. 2, pp. 155-174(20), Blackwell Publishing Ltd
159
Brovelli M. A. and Cannata M., 2004, Digital Terrain model
160
reconstruction in urban areas from airborne laser scanning data: the
161
method and an example for Pavia (Northern Italy). Computers and
162
Geosciences 30, pp.325-331
164
Brovelli M. A e Longoni U.M., 2003, Software per il filtraggio di
165
dati LIDAR, Rivista dell'Agenzia del Territorio, n. 3-2003, pp. 11-22
168
Antolin R. and Brovelli M.A., 2007, LiDAR data Filtering with GRASS GIS for the
169
Determination of Digital Terrain Models. Proceedings of Jornadas de SIG Libre,
170
Girona, España. CD ISBN: 978-84-690-3886-9 <br>
172
<p><i>Last changed: $Date: 2012-12-27 18:22:59 +0100 (Thu, 27 Dec 2012) $</i>