Abstract.Wavelet expansion has been demonstrated to be suitable for the representation of spatial functions. Here we propose the so-called B-spline wavelets to represent spatial time-series of GPS-derived global ionosphere maps (GIMs) of the vertical total electron content (TEC) from the Earth’s surface to the mean altitudes of GPS satellites, over Japan. The scalar-valued B-spline wavelets can be defined in a two-dimensional, but not necessarily planar, domain. Generated by a sequence of knots, different degrees of B-splines can be implemented: degree 1 represents the Haar wavelet; degree 2, the linear B-spline wavelet, or degree 4, the cubic B-spline wavelet. A non-uniform version of these wavelets allows us to handle data on a bounded domain without any edge effects. B-splines are easily extended with great computational efficiency to domains of arbitrary dimensions, while preserving their properties. This generalization employs tensor products of B-splines, defined as linear superposition of products of univariate B-splines in different directions. The data and model may be identical at the locations of the data points if the number of wavelet coefficients is equal to the number of grid points. In addition, data compression is made efficient by eliminating the wavelet coefficients with negligible magnitudes, thereby reducing the observational noise. We applied the developed methodology to the representation of the spatial and temporal variations of GIM from an extremely dense GPS network, the GPS Earth Observation Network (GEONET) in Japan. Since the sampling of the TEC is registered regularly in time, we use a two-dimensional B-spline wavelet representation in space and a one-dimensional spline interpolation in time. Over the Japan region, the B-spline wavelet method can overcome the problem of bias for the spherical harmonic model at the boundary, caused by the non-compact support. The hierarchical decomposition not only allows an inexpensive calculation, but also separates visualisation at different levels of detail. Each level corresponds to a certain spatial frequency band, leading to a detection of structures and enhancement in the ionosphere at different resolutions.
[1]
Anthony J. Mannucci,et al.
Subdaily northern hemisphere ionospheric maps using an extensive network of GPS receivers
,
1995
.
[2]
E. J. Stollnitz,et al.
Wavelets for Computer Graphics : A Primer
,
1994
.
[3]
C. Chui,et al.
On compactly supported spline wavelets and a duality principle
,
1992
.
[4]
R. Mautz,et al.
Toward a Spherical Pseudo-Wavelet Basis for Geodetic Applications
,
2003
.
[5]
Tom Lyche,et al.
A Multiresolution Tensor Spline Method for Fitting Functions on the Sphere
,
2000,
SIAM J. Sci. Comput..
[6]
Leos Mervart,et al.
Global and regional ionosphere models using the GPS double difference phase observable
,
1996
.
[7]
E. Quak,et al.
Decomposition and Reconstruction Algorithms for Spline Wavelets on a Bounded Interval
,
1994
.
[8]
C. Chui,et al.
Wavelets on a Bounded Interval
,
1992
.
[9]
Koji Matsumoto,et al.
Regional ionosphere map over Japanese Islands
,
2002
.
[10]
P. H. Salamonowicz.
A Wavelet Based Gravity Model with an Application to the Evaluation of Stokes’ Integral
,
2001
.
[11]
Karl-Rudolf Koch,et al.
Parameter estimation and hypothesis testing in linear models
,
1988
.
[12]
Takuya Tsugawa,et al.
A new technique for mapping of total electron content using GPS network in Japan
,
2002
.
[13]
Shoichiro Fukao,et al.
High resolution mapping of TEC perturbations with the GSI GPS Network over Japan
,
1998
.