Elsevier

Remote Sensing of Environment

Volume 141, 5 February 2014, Pages 144-148
Remote Sensing of Environment

SST spatial anisotropic covariances from METOP-AVHRR data

https://doi.org/10.1016/j.rse.2013.10.024Get rights and content

Highlights

  • Completely observation-based approach

  • Anisotropy of the SST spatial variability is fitted by oriented ellipsoids.

  • Regions with high isotropic or anisotropic behavior are highlighted.

  • Regions with high measurement errors of the METOP-AVHRR are highlighted.

  • The spatial variability parameters for the global Ocean are available online.

Abstract

The Advanced Very High Resolution Radiometer (AVHRR) instrument on-board the METOP satellite is designed to provide very accurate measurements of Sea Surface Temperature (SST). In this work, using one year of METOP-AVHRR data and a geostatistical approach, we characterize the spatial anisotropy and non-stationarity of the SST variability using oriented ellipsoids. The method is also able to separate the true SST variability from the artificial error introduced by the METOP-AVHRR sensor. These spatial parameters are then used for producing variability atlases (available on-line) over the whole ocean.

Introduction

Sea Surface Temperature (SST) is a key geophysical variable for many applications (see e.g. Donlon et al., 2002 and reference therein). For instance, global-scale, complete and realistic SST fields are key input data in the Ocean General Circulation Models (OGCMs). One may also cite the use of SST data to track, characterize and reconstruct mesoscale dynamics (cf. Isern-Fontanet et al., 2006, Klein et al., 2009, Tandeo et al., 2013). In this respect, satellite sensing data provide invaluable global-scale data with relatively high spatial and temporal resolutions. Such data are however vulnerable to various atmospheric contaminations (e.g., cloud cover, aerosols, heavy rains) and they may present large missing data rates. Therefore, spatial and temporal interpolation techniques are necessary to reconstruct the hidden information (see e.g. Reynolds et al., 2007). Understanding and modeling of mesoscale SST variability appear crucial to provide interpolation priors.

The temporal variability of the SST has been investigated for short time scales from in situ data (cf. Wainer, Clauzet, Servain, & Soares, 2003) and from remote sensing sources (cf. Tandeo, Ailliot, & Autret, 2011) but, its spatial variability, especially spatial anisotropy and non-stationarity, remains poorly explored. Whereas regional studies addressed the isotropic spatial variability of the SST (e.g. Gohin & Langlois, 1993 in the bay of Biscay), evidences for local spatial anisotropies have been reported for SST fields in some studies (e.g. Park & Chung, 1999 in the Sea of Japan) as well as for other geophysical variables such as ocean color (cf. Doney, Glover, McCue, & Fuentes, 2003). However, to our knowledge, no study has explored thoroughly at a global-scale these mesoscale spatial SST variabilities. This study relies on geostatistic tools to address these issues from the analysis of the Advanced Very High Resolution Radiometer (AVHRR) instrument on-board the METOP-A satellite (cf. Le Borgne et al., 2007, O&SI SAF Project Team, 2013). Using compact parametric model of the spatial covariance of the SST fields and the associated model fitting scheme, we produce observation-based atlases of the mesoscale spatial variability of the SST for the global Ocean, in terms of local anisotropy, spatial scales and residual variances. These atlases provide new insights on global and regional mesoscale ocean dynamics and new priors for missing data interpolation.

This paper is organized as follows. The METOP-AVHRR SST data and the geostatistical methodology are introduced in Section 2. We discuss the results obtained in the Malvinas current and the global Ocean in Section 3. Section 4 further discusses our key contributions and prospects for future work.

Section snippets

METOP-AVHRR data

METOP-A satellite has a high resolution coverage of the world Ocean. The data are delivered in satellite projection at full resolution and remapped onto a regular 0.05° × 0.05° global grid every 12 h (see O&SI SAF Project Team, 2013 for more details). METOP-AVHRR is one of the most efficient SST data source due to the high quality measurements and the number of observations (cf. Le Borgne, Legendre, & Marsouin (2007)). In this study, we only use night-time data to avoid the variability due to the

Results

The estimated parameters σ02 (nugget), σ2 (sill), ϕ (anisotropic angle), Lmin (minimum range) and Lmax (maximum range) are discussed in this section. They are computed for a 1° × 1° gridover the global Ocean. Locally, we can suppose that the spatial variability is second-order stationary. The computation of the empirical semivariograms proceeds as follows. For a given 1° × 1° area, we first compute daily empirical semivariograms: we sort all locations within the considered area and retrieve the

Conclusion

In this paper, we have presented a geostatistic method to characterize the non-stationarity of the spatial variability of the SST at mesoscale from 2008 METOP-AVHRR data. We chose an exponential model with an anisotropic parameterization to take into account the differences of the SST spatial variability along the directions. Five parameters were estimated in the global Ocean. They are available online as supplementary material. Among them, we find the stationary variance, the anisotropy

Acknowledgment

The data from the EUMETSAT Satellite Application Facility on Ocean and Sea Ice are accessible through the SAF's homepage: http://www.osi-saf.org/. We are grateful to Dr. Pierre Le Borgne for his expertise and valuable comments on this work.

References (17)

  • N.A.C. Cressie

    Statistics for spatial data

    (1993)
  • P.J. Diggle et al.

    Model-based geostatistics

    (2007)
  • S.C. Doney et al.

    Mesoscale variability of SeaWiFS satellite ocean color: Global patterns and spatial scales

    Journal of Geophysical Research

    (2003)
  • C.J. Donlon et al.

    Toward improved validation of satellite sea surface skin temperature measurements for climate research

    Journal of Climate

    (2002)
  • G.R. Foltz et al.

    Impact of Saharan dust on tropical North Atlantic SST

    Journal of Climate

    (2008)
  • F. Gohin et al.

    Using geostatistics to merge in situ measurements and remotely-sensed observations of sea surface temperature

    International Journal of Remote Sensing

    (1993)
  • J. Isern-Fontanet et al.

    Potential use of microwave sea surface temperatures for the estimation of ocean currents

    Geophysical Research Letters

    (2006)
  • P. Klein et al.

    Diagnosis of vertical velocities in the upper ocean from high resolution sea surface height

    Geophysical Research Letters

    (2009)
There are more references available in the full text version of this article.

Cited by (22)

  • Submarine groundwater discharge detection through remote sensing: An application of Landsat 7 and 8 in Hawaiʻi and Ireland

    2022, Remote Sensing of Environment
    Citation Excerpt :

    Both the derivative analysis and the AD methods work on a pixel-wise manner, making them independent of scene area. On the contrary, the definition of the SST intervals depends on the area of the scene, since SST range and variability are a function of area (Tandeo et al., 2014) (among other factors). However, as discussed below, local SST anomalies may exist outside of the coldest SST interval.

  • Application of Satellite Data on Water Pollution: An Intensive Review

    2022, Journal of Environmental Informatics Letters
View all citing articles on Scopus
1

Both authors equally contributed to this work.

View full text