Spectral analysis using orthonormal functions with a case study on the sea surface topography

Chein-way Hwang*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

22 Scopus citations


b Spherical harmonics are orthonormalized using the Gram‐Schmidt process in a function space. The problem of linear dependence of spherical harmonics over the oceans is studied using the Gram matrices and consequently three sets of orthonormal (ON) functions have been constructed. For the process an efficient formula for computing inner products of spherical harmonics has been developed. Important spectral properties of the ON functions are addressed. The ON functions may be used for representing the sea surface topography (SST) in the analysis of satellite altimeter data. The geoid error can be transformed to a representation by the ON functions and hence the comparison of powers of the geoid error and the SST signal only over the oceans is possible, leading to a better way of determining the cut‐off frequency of the SST in the simultaneous solution using satellite altimeter data. As a case study, the modified Levitus SST is expanded into the ON functions. The results show that 99.90 per cent of that signal's energy is contained within degree 24 of the orthonormal functions. Such expansions also render better spectral behaviour of oceanic signals as compared to that from spherical harmonic expansions. The study shows that these generalized Fourier functions are suitable for spectral analyses of oceanic signals and they can be applied to future altimetric mission where the geoid and the SST are to be recovered.

Original languageEnglish
Pages (from-to)1148-1160
Number of pages13
JournalGeophysical Journal International
Issue number3
StatePublished - 1 Jan 1993


  • Key words:
  • SST
  • geoid
  • orthonormal functions
  • spectrum
  • spherical harmonic


Dive into the research topics of 'Spectral analysis using orthonormal functions with a case study on the sea surface topography'. Together they form a unique fingerprint.

Cite this