-
Notifications
You must be signed in to change notification settings - Fork 103
/
shbiaskmask.doc
72 lines (63 loc) · 3.45 KB
/
shbiaskmask.doc
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
Calculate the multitaper (cross-)power spectrum expectation of a function
localized by arbitrary windows derived from a mask.
Usage
-----
outcspectra = SHBiasK (tapers, incspectra, [lwin, k, ldata, taper_wt, save_cg])
Returns
-------
outcspectra : float, dimension (ldata+lwin+1)
The expectation of the multitaper localized power spectrum.
Parameters
----------
tapers : float, dimension ((lwinin+1)**2, kin)
The spherical harmonic coefficients of the localization windows generated by
a call to SHReturnTapersMap. The coefficients in each column are ordered
according to the convention in SHCilmToVector.
incspectra : float, dimension (ldatain+1)
The global unwindowed power spectrum.
lwin : optional, integer, default = lwinin
The spherical harmonic bandwidth of the localizing windows.
k : optional, integer, default = kin
The number of localization windows to use. Only the first k columns of
tapers will be employed, which corresponds to the best-concentrated windows.
ldata : optional, integer, default = ldatain
The maximum degree of the global unwindowed power spectrum.
taper_wt : optional, float, dimension (kin), default = -1
The weights to apply to each individual windowed specral estimate. The
weights must sum to unity, and the default specifies that taper weights are
not used.
save_cg : optional, integer, default = 0
If set equal to 1, the Clebsch-Gordon coefficients will be precomputed and
saved for future use (if lwin or ldata change, these will be recomputed). To
deallocate the saved memory, set this parameter equal to 1. If set equal to
0 (default), the Clebsch-Gordon coefficients will be recomputed for each
call.
Description
-----------
SHBiasKMask will calculate the multitaper (cross-)power spectrum expectation of
a function multiplied by the k best-concentrated localization windows derived
from an arbitrary mask. This is given by equation 36 of Wieczorek and Simons
(2005) (see also eq. 2.11 of Wieczorek and Simons 2007). In contrast to SHBias,
which takes as input the power spectrum of a single localizing window, this
routine expects as input a matrix containing the spherical harmonic coefficients
of the localizing windows. These can be generated by a call to SHReturnTapersMap
and the coefficients in each column are ordered according to the convention in
SHCilmToVector.
It is assumed implicitly that the power spectrum of inspectrum is zero beyond
degree ldata. If this is not the case, the ouput power spectrum should be
considered valid only for the degrees up to and including ldata - lwin.
The default is to apply equal weights to each individual windowed estimate of
the spectrum, but this can be modified by specifying the weights in the optional
argument taper_wt. The weights must sum to unity. If this routine is to be
called several times using the same values of lwin and ldata, then the Clebsch-
Gordon coefficients can be precomputed and saved by setting the optional
parameter save_cg equal to 1.
References
----------
Wieczorek, M. A. and F. J. Simons, Minimum-variance multitaper spectral
estimation on the sphere, J. Fourier Anal. Appl., 13, 665-692,
doi:10.1007/s00041-006-6904-1, 2007.
Simons, F. J., F. A. Dahlen and M. A. Wieczorek, Spatiospectral concentration on
a sphere, SIAM Review, 48, 504-536, doi:10.1137/S0036144504445765, 2006.
Wieczorek, M. A. and F. J. Simons, Localized spectral analysis on the sphere,
Geophys. J. Int., 162, 655-675, doi:10.1111/j.1365-246X.2005.02687.x, 2005.