JPL MUR SST

Multi-Resolution Analysis:

See PO.DAAC MEaSUREs MUR SST page for more updated details.

Satellite-based SST data sets have drastically different resolutions. For example, the infra-red (IR) type sensor can have a very high resolution down to 1 km in horizontal distance) but the micro-wave (MW) sensor data are typically much coarser at 25 km resolution. All data are usually sampled in irregular patterns under the satellite orbits. On top of all these, there are much data-voids (holes in data coverage) due to contaminations by clouds, aerosols, and land (depending on the sensor).

To deal with these sampling issues objectively, we use a technique called Multi-Resolution Variational Analysis (MRVA). MRVA is a statistical interpolation method based on wavelet decomposition called the "multi-resolution analysis". MRVA uses an orthonormal (energy conserving) transformation of a signal into a scale space, similar to a signal transformation into the Fourier wavenumber space. The notion of "scale" in MRA is similar to the wavelength in Fourier transform; however, an important difference is that the scale is space-specific (localized) while the wavenumber mathematically has a global extent. So with MRVA, smoothness (spectral contents) of the SST map can be controlled without degrading local representativeness.

MUR uses the Battle-Lemarie wavelet basis, depicted below over three consecutive scales. Technical details are found in Chin et al 1998, but please note that MUR uses NO statistical synthesis of the wavelet coefficients. All coefficients are derived from the SST data.

References

Chin, T.M., Milliff, R.F., and Large, W.G., (1998). Basin-scale, high-wavenumber sea surface wind fields from a multiresolution analysis of scatterometer data. Journal of Atmospheric and Oceanic Technology, 15: 741-763. [link]


		Multi-Resolution Analysis: See PO.DAAC MEaSUREs MUR SST page for more updated details. Satellite-based SST data sets have drastically different resolutions. For example, the infra-red (IR) type sensor can have a very high resolution down to 1 km in horizontal distance) but the micro-wave (MW) sensor data are typically much coarser at 25 km resolution. All data are usually sampled in irregular patterns under the satellite orbits. On top of all these, there are much data-voids (holes in data coverage) due to contaminations by clouds, aerosols, and land (depending on the sensor). To deal with these sampling issues objectively, we use a technique called Multi-Resolution Variational Analysis (MRVA). MRVA is a statistical interpolation method based on wavelet decomposition called the "multi-resolution analysis". MRVA uses an orthonormal (energy conserving) transformation of a signal into a scale space, similar to a signal transformation into the Fourier wavenumber space. The notion of "scale" in MRA is similar to the wavelength in Fourier transform; however, an important difference is that the scale is space-specific (localized) while the wavenumber mathematically has a global extent. So with MRVA, smoothness (spectral contents) of the SST map can be controlled without degrading local representativeness. MUR uses the Battle-Lemarie wavelet basis, depicted below over three consecutive scales. Technical details are found in Chin et al 1998, but please note that MUR uses NO statistical synthesis of the wavelet coefficients. All coefficients are derived from the SST data. References Chin, T.M., Milliff, R.F., and Large, W.G., (1998). Basin-scale, high-wavenumber sea surface wind fields from a multiresolution analysis of scatterometer data. Journal of Atmospheric and Oceanic Technology, 15: 741-763. [link]