1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
|
<HTML>
<HEAD>
<TITLE>trend2d</TITLE>
</HEAD>
<BODY>
<H1>trend2d</H1>
<HR>
<PRE>
<!-- Manpage converted by man2html 3.0.1 -->
trend2d - Fit a [weighted] [robust] polynomial model for z
= f(x,y) to xyz[w] data.
</PRE>
<H2>SYNOPSIS</H2><PRE>
<B>trend2d</B> <B>-F</B><I><xyzmrw></I> <B>-N</B><I>n</I><B>_</B><I>model</I>[<B>r</B>] [ <I>xyz[w]file</I> ] [ <B>-C</B><I>condi</I>
<I>tion</I><B>_</B><I>#</I> ] [ <B>-H</B>[<I>nrec</I>] ][ <B>-I</B>[<I>confidence</I><B>_</B><I>level</I>] ] [ <B>-V</B> ] [ <B>-W</B>
] [ <B>-:</B> ] [ <B>-bi</B>[<B>s</B>][<I>n</I>] ] [ <B>-bo</B>[<B>s</B>] ]
</PRE>
<H2>DESCRIPTION</H2><PRE>
<B>trend2d</B> reads x,y,z [and w] values from the first three
[four] columns on standard input [or <I>xyz[w]file</I>] and fits
a regression model z = f(x,y) + e by [weighted] least
squares. The fit may be made robust by iterative reweight
ing of the data. The user may also search for the number
of terms in f(x,y) which significantly reduce the variance
in z. n_model may be in [1,10] to fit a model of the fol
lowing form (similar to grdtrend):
m1 + m2*x + m3*y + m4*x*y + m5*x*x + m6*y*y + m7*x*x*x +
m8*x*x*y + m9*x*y*y + m10*y*y*y.
The user must specify <B>-N</B><I>n</I><B>_</B><I>model</I>, the number of model
parameters to use; thus, <B>-N</B><I>4</I> fits a bilinear trend, <B>-N</B><I>6</I> a
quadratic <A HREF="surface.html">surface</A>, and so on. Optionally, append <B>r</B> to per
form a robust fit. In this case, the program will itera
tively reweight the data based on a robust scale estimate,
in order to converge to a solution insensitive to out
liers. This may be handy when separating a "regional"
field from a "residual" which should have non-zero mean,
such as a local mountain on a regional surface.
<B>-F</B> Specify up to six letters from the set {x y z m r
w} in any order to create columns of ASCII [or
binary] output. x = x, y = y, z = z, m = model
f(x,y), r = residual z - m, w = weight used in fit
ting.
<B>-N</B> Specify the number of terms in the model, <I>n</I><B>_</B><I>model</I>,
and append <B>r</B> to do a robust fit. E.g., a robust
bilinear model is <B>-N</B><I>4</I><B>r</B>.
</PRE>
<H2>OPTIONS</H2><PRE>
<I>xyz[w]file</I>
ASCII [or binary, see <B>-b</B>] file containing x,y,z [w]
values in the first 3 [4] columns. If no file is
specified, <B>trend2d</B> will read from standard input.
<B>-C</B> Set the maximum allowed condition number for the
matrix solution. <B>trend2d</B> fits a damped least
squares model, retaining only that part of the
eigenvalue spectrum such that the ratio of the
largest eigenvalue to the smallest eigenvalue is
header records can be changed by editing your .gmt
defaults file. If used, <B><A HREF="GMT.html">GMT</A></B> default is 1 header
record.
<B>-I</B> Iteratively increase the number of model parame
ters, starting at one, until <I>n</I><B>_</B><I>model</I> is reached or
the reduction in variance of the model is not sig
nificant at the <I>confidence</I><B>_</B><I>level</I> level. You may set
<B>-I</B> only, without an attached number; in this case
the fit will be iterative with a default confidence
level of 0.51. Or choose your own level between 0
and 1. See remarks section.
<B>-V</B> Selects verbose mode, which will send progress
reports to stderr [Default runs "silently"].
<B>-W</B> Weights are supplied in input column 4. Do a
weighted least squares fit [or start with these
weights when doing the iterative robust fit].
[Default reads only the first 3 columns.]
<B>-:</B> Toggles between (longitude,latitude) and (lati
tude,longitude) input/output. [Default is (longi
tude,latitude)]. Applies to geographic coordinates
only.
<B>-bi</B> Selects binary input. Append <B>s</B> for single precision
[Default is double]. Append <I>n</I> for the number of
columns in the binary file(s). [Default is 3 (or 4
if <B>-W</B> is set) input columns].
<B>-bo</B> Selects binary output. Append <B>s</B> for single preci
sion [Default is double].
</PRE>
<H2>REMARKS</H2><PRE>
The domain of x and y will be shifted and scaled to [-1,
1] and the basis functions are built from Chebyshev poly
nomials. These have a numerical advantage in the form of
the matrix which must be inverted and allow more accurate
solutions. In many applications of <B>trend2d</B> the user has
data located approximately along a line in the x,y plane
which makes an angle with the x axis (such as data col
lected along a road or ship track). In this case the accu
racy could be improved by a rotation of the x,y axes.
<B>trend2d</B> does not search for such a rotation; instead, it
may find that the matrix problem has deficient rank. How
ever, the solution is computed using the generalized
inverse and should still work out OK. The user should
check the results graphically if <B>trend2d</B> shows deficient
rank. NOTE: The model parameters listed with <B>-V</B> are Cheby
shev coefficients; they are not numerically equivalent to
the m#s in the equation described above. The description
The <B>-N</B><I>n</I><B>_</B><I>model</I><B>r</B> (robust) and <B>-I</B> (iterative) options evalu
ate the significance of the improvement in model misfit
Chi-Squared by an F test. The default confidence limit is
set at 0.51; it can be changed with the <B>-I</B> option. The
user may be surprised to find that in most cases the
reduction in variance achieved by increasing the number of
terms in a model is not significant at a very high degree
of confidence. For example, with 120 degrees of freedom,
Chi-Squared must decrease by 26% or more to be significant
at the 95% confidence level. If you want to keep iterating
as long as Chi-Squared is decreasing, set <I>confidence</I><B>_</B><I>level</I>
to zero.
A low confidence limit (such as the default value of 0.51)
is needed to make the robust method work. This method
iteratively reweights the data to reduce the influence of
outliers. The weight is based on the Median Absolute Devi
ation and a formula from Huber [1964], and is 95% effi
cient when the model residuals have an outlier-free normal
distribution. This means that the influence of outliers is
reduced only slightly at each iteration; consequently the
reduction in Chi-Squared is not very significant. If the
procedure needs a few iterations to successfully attenuate
their effect, the significance level of the F test must be
kept low.
</PRE>
<H2>EXAMPLES</H2><PRE>
To remove a planar trend from data.xyz by ordinary least
squares, try:
<B>trend2d</B> data.xyz <B>-F</B>xyr <B>-N</B>2 > detrended_data.xyz
To make the above planar trend robust with respect to out
liers, try:
<B>trend2d</B> data.xzy <B>-F</B>xyr <B>-N</B>2<B>r</B> > detrended_data.xyz
To find out how many terms (up to 10) in a robust inter
polant are significant in fitting data.xyz, try:
<B>trend2d</B> data.xyz <B>-N</B>10<B>r</B> <B>-I</B> <B>-V</B>
</PRE>
<H2>SEE ALSO</H2><PRE>
<I>gmt</I>(l), <I><A HREF="grdtrend.html">grdtrend</A></I>(l), <I><A HREF="trend1d.html">trend1d</A></I>(l)
</PRE>
<H2>REFERENCES</H2><PRE>
Huber, P. J., 1964, Robust estimation of a location param
eter, <I>Ann.</I> <I>Math.</I> <I>Stat.,</I> <I>35,</I> 73-101.
Menke, W., 1989, Geophysical Data Analysis: Discrete
Inverse Theory, Revised Edition, Academic Press, San
</PRE>
<HR>
<ADDRESS>
Man(1) output converted with
<a href="http://www.oac.uci.edu/indiv/ehood/man2html.html">man2html</a>
</ADDRESS>
</BODY>
</HTML>
<body bgcolor="#ffffff">
|
