The amplitude of the variability associated with dynamic mesoscale phenomena is most often greater than that associated with seasonal variability. The role of these phenomena is of the utmost importance as they modify the general circulation of the water masses, and thus, potentially, the circulation in the coastal zone too. But in situ observations at mesoscale require fine spatiotemporal sampling, requiring much effort. Under these conditions, however, NOAA/AVHRR thermal satellite imagery is an extremely efficient tool, as it routinely provides information on a spatial domain over thousands of kilometres, and can generate high-resolution long-term time series. Providing somne precautions are taken, thermal signatures can be interpreted in terms of dynamical structures and associated currents. The use of this satellite imagery within the operation ELISA (1997-1998) provides an opportunity to review the potential limitations to an automatic recognition (detection and tracking) of such mesoscale structures.
We consider the statistical analysis of high frequency sampled time series in nearshore waters of the Reunion island, located in the Indian Ocean 700 km east of Madagascar. We focus particularly on automatic measurements of current data sets at four different stations in the island. As this entire dataset is characterized by non-linearity and non-stationarity, two time-frequency decomposition methods, namely wavelets and Hilbert-Huang Transformation (HHT), are applied for the analysis of the entire dataset. Using the Continuous Wavelet Transform (CWT), some properties of the currents time series are determined. Then, the inertial wave and several low-frequency tidal waves are identified by the application of the Empirical Mode Decomposition (EMD) and the associated Hilbert spectral analysis (resp. HHT). Both wavelet and HHT approaches provide very useful information about the characteristics of currents measurements, although a poor low-frequency resolution is discerned for the Morlet wavelet spectra. Furthermore, the authors investigate the cross-correlations between currents at the different stations. Wavelet coherence and EMD based Time Dependent Intrinsic Correlation (TDIC) analyses are applied to consider the correlation between two nonstationary time series. By TDIC analysis, it was concluded that the high frequency modes have small correlation; whereas the trends are perfectly correlated. The results obtained by wavelet coherence are very similar, thus confirming that both approaches could be used for identification of main properties of marine environmental time series. The methodologies presented in this paper are general and thus can be applied on other time series from the environmental and oceanic sciences, where time series are complex with fluctuations over a large range of different spatial and temporal scales, from seconds to thousands of years. Keywords: Continuous wavelet transform, cross-correlation, cross-wavelet, empirical mode decomposition, Hilbert-Huang transform, Hilbert spectral anal-ysis, Morlet wavelet, stationarity, time dependent intrinsic correlation, time series, wavelets, wavelet coherence.
In marine sciences, time series are often nonlinear and nonstationary. Adequate and specific methods are needed to analyze such series. In this letter, an application of the empirical mode decomposition method (EMD) associated to the Hilbert spectral analysis (HSA) is presented. Furthermore, EMD-based time-dependent intrinsic correlation (TDIC) analysis is applied to consider the correlation between two nonstationary time series. Four temperature time series obtained from automatic measurements in nearshore waters of the Réunion island are considered, recorded every 10 min from July 2011 to January 2012. The application of the EMD on these series and the estimation of their power spectra using the HSA are illustrated. The authors identify low-frequency tidal waves and display the pattern of correlations at different scales and different locations. By TDIC analysis, it was concluded that the high-frequency modes have small correlation, whereas the trends are perfectly correlated.
On the French continental shelf of the Bay of Biscay the variability of the surface salinity distribution has been mainly investigated at the seasonal and inter-annual scales.Here, new mesoscale features such as lower-salinity lenses observed in model results are investigated by hydrological measurements acquired during 8 cruises (1997)(1998)(1999)(2000).These lenses are 50-80 km wide and ~30 m thick and occur during westerly to northerly wind events that push offshore the less saline water of river plumes.These water masses detached from the coast are replaced with upwelled saltier water at the coast, so coastal upwelling is often observed at the same time along Landes and southern Brittany coasts.We show that in addition to the influence of seasonal and inter-annual variability of the wind and river outflows, short term meteorological variability may drive mesoscale structures on this continental shelf.
Abstract Five oceanographic cruises were organized in the Sardinian Sea and Channel in May 2000, March 2001, September 2001, May 2002, and November 2002 to study the characterization of the water masses, Atlantic Water (AW) and Winter Intermediate Water (WIW), and their mesoscale variability. In the Sardinian Channel, an Algerian anticyclonic Eddy (AE) was observed in May 2000, along the Tunisian coast. This induced a greater minimum salinity in a wider and deeper layer than in November 2002, when no AE was observed. Some WIW was observed below it; nevertheless, no link could be established between AEs and WIW occurrences. In the Sardinian Sea, two AEs were observed during spring 2000, and a further two during spring 2002. One AE strongly influenced shelf circulation, in contrast to the other three that were off the continental slope. In the same area, during the end of September 2001, a vertical salinity inversion occurred in the first 30–50 m of depth over the whole sampling field, and a W–NW wind induced a coastal upwelling over the western Sardinian coast (south of 41° N). This upwelling increased the salinity from ∼20 to 30 m below the surface to the surface and, thereby led to a lower salinity close to the coast than offshore. This was in contrast to a classical upwelling. Consequently, in the Sardinian Sea, the general circulation, mainly driven by AEs, can meet the coastal wind-driven circulation. Keywords: SardiniaAWWIWAnticyclonic eddymesoscale Acknowledgements The authors would especially like to thank all the people who sailed on the R/V Urania for their great contribution during the five MedGOOS cruises, the German Aerospace Centre DLR (http://eoweb.dlr.de) for the infrared images and Elena Mauri and Pierre-Marie Poulain from the Remote Sensing Group (SIRE) of OGS (Trieste, Italy) for providing the processed satellite images. This work was financially supported by: the Italian MIUR in the frame of the project SIMBIOS (SIstema per lo studio del Mare con Boa Integrata OffShore—Operative Programme of the Marine Environment Plan, Cluster C10, Project n. 13—D.n. 778.RIC) and the EU Marie Curie Host Fellowship of the Human Potential programme, project ODASS (contract n. HPMD-CT-2001-00075). This work was developed in the framework of the EuroGOOS and MedGOOS strategies and in synergy with the EU project MAMA (contract no. EVR1-CT-2001-20010 MAMA).
Dans l’environnement marin, la temperature, la turbidite et la salinite sont des quantites importantes pour etudier l’ecosysteme. Beaucoup de champs montrent la haute variabilite pour une large gamme d'echelles spatiale et temporelle. Les series temporelles sont souvent non-lineaires et non-stationnaires. Les auteurs considerent ici les fluctuations du niveau de la mer, la salinite, la turbidite et la temperature enregistres par le systeme MAREL Carnot de Boulogne-sur-Mer (France), qui est une bouee amarree equipee d'appareils de mesure physicochimiques, qui fonctionnent dans des conditions continues et autonomes. Pour realiser des analyses statistiques et spectrales adequates, il est necessaire de connaitre la nature de la serie temporelle consideree. Pour cette raison, la stationnarite de la serie et l'occurrence de racine d'unite sont abordees avec les Tests Augmentes de Dickey- Fuller. A part pour le niveau de la mer, l'analyse harmonique n'est pas pertinente. Le grand nombre de donnees fournies par les capteurs permet l'evaluation de l’analyse spectrale de Fourier, pour considerer les frequences dominantes associees a la dynamique. Differents spectres de puissance montrent une variabilite complexe et revelent une influence de facteurs exogenes tels que les marees. Cependant, l'analyse spectrale classique, a savoir la methode Blackman-Tukey, exige des donnees regulierement espacees. L'interpolation de la serie temporelle introduit de nombreux artefacts aux donnees. L'algorithme de Lomb-Scargle, adapte aux donnees inegalement espacees, est utilise comme alternative. Les limites de la methode sont aussi exposees. On a trouve qu'au-dela de 50 % de mesures manquantes, peu de frequences significatives sont detectees, plusieurs saisonnalites ne sont plus visibles, et meme une gamme entiere de haute frequence disparait progressivement.
Abstract. Submesoscale processes have a determinant role in the dynamics of oceans by transporting momentum, heat, mass, and particles. Furthermore, they can define niches where different phytoplankton species flourish and accumulate not only by nutrient provisioning but also by modifying the water column structure or active gathering through advection. In coastal areas, however, submesoscale oceanic processes act together with coastal ones, and their effect on phytoplankton distribution is not straightforward. The present study brings the relevance of hydrodynamic variables, such as vorticity, into consideration in the study of phytoplankton distribution, via the analysis of in situ and remote multidisciplinary data. In situ data were obtained during the ETOILE oceanographic cruise, which surveyed the Capbreton Canyon area in the southeastern part of the Bay of Biscay in early August 2017. The main objective of this cruise was to describe the link between the occurrence and distribution of phytoplankton spectral groups and mesoscale to submesoscale ocean processes. In situ discrete hydrographic measurements and multi-spectral chlorophyll a (chl a) fluorescence profiles were obtained in selected stations, while temperature, conductivity, and in vivo chl a fluorescence were also continuously recorded at the surface. On top of these data, remote sensing data available for this area, such as high-frequency radar and satellite data, were also processed and analysed. From the joint analysis of these observations, we discuss the relative importance and effects of several environmental factors on phytoplankton spectral group distribution above and below the pycnocline and at the deep chlorophyll maximum (DCM) by performing a set of generalized additive models (GAMs). Overall, salinity is the most important parameter modulating not only total chl a but also the contribution of the two dominant spectral groups of phytoplankton, brown and green algae groups. However, at the DCM, among the measured variables, vorticity is the main modulating environmental factor for phytoplankton distribution and explains 19.30 % of the variance. Since the observed distribution of chl a within the DCM cannot be statistically explained without the vorticity, this research sheds light on the impact of the dynamic variables in the distribution of spectral groups at high spatial resolution.