Limits...
Defining Mediterranean and Black Sea biogeochemical subprovinces and synthetic ocean indicators using mesoscale oceanographic features.

Nieblas AE, Drushka K, Reygondeau G, Rossi V, Demarcq H, Dubroca L, Bonhommeau S - PLoS ONE (2014)

Bottom Line: Principal components analysis is then performed on the oceanographic variables to define integrative indices to monitor the environmental changes within each resultant subprovince at monthly resolutions.The first axis of the principal component analysis is explained primarily by classical ocean features and the second axis is explained by mesoscale features.Biogeochemical subprovinces identified by the present study can be useful within the European management framework as an objective geographical framework of the Mediterranean and Black Seas, and the synthetic ocean indicators developed here can be used to monitor variability and long-term change.

View Article: PubMed Central - PubMed

Affiliation: Unité Mixte Recherche Ecosystèmes Marins Exploités 212, Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER), Sète, France.

ABSTRACT
The Mediterranean and Black Seas are semi-enclosed basins characterized by high environmental variability and growing anthropogenic pressure. This has led to an increasing need for a bioregionalization of the oceanic environment at local and regional scales that can be used for managerial applications as a geographical reference. We aim to identify biogeochemical subprovinces within this domain, and develop synthetic indices of the key oceanographic dynamics of each subprovince to quantify baselines from which to assess variability and change. To do this, we compile a data set of 101 months (2002-2010) of a variety of both "classical" (i.e., sea surface temperature, surface chlorophyll-a, and bathymetry) and "mesoscale" (i.e., eddy kinetic energy, finite-size Lyapunov exponents, and surface frontal gradients) ocean features that we use to characterize the surface ocean variability. We employ a k-means clustering algorithm to objectively define biogeochemical subprovinces based on classical features, and, for the first time, on mesoscale features, and on a combination of both classical and mesoscale features. Principal components analysis is then performed on the oceanographic variables to define integrative indices to monitor the environmental changes within each resultant subprovince at monthly resolutions. Using both the classical and mesoscale features, we find five biogeochemical subprovinces for the Mediterranean and Black Seas. Interestingly, the use of mesoscale variables contributes highly in the delineation of the open ocean. The first axis of the principal component analysis is explained primarily by classical ocean features and the second axis is explained by mesoscale features. Biogeochemical subprovinces identified by the present study can be useful within the European management framework as an objective geographical framework of the Mediterranean and Black Seas, and the synthetic ocean indicators developed here can be used to monitor variability and long-term change.

Show MeSH
Biogeochemical subprovinces of the Mediterranean and Black Seas.Subprovinces for the (a) “classical”, (b) “mesoscale”, and (c) “full” multivariate arrays using a 5% threshold for the explained sum of squares to define the optimal number of subprovinces (see text).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4216069&req=5

pone-0111251-g002: Biogeochemical subprovinces of the Mediterranean and Black Seas.Subprovinces for the (a) “classical”, (b) “mesoscale”, and (c) “full” multivariate arrays using a 5% threshold for the explained sum of squares to define the optimal number of subprovinces (see text).

Mentions: Multivariate arrays were created from time-averages of the monthly scaled oceanographic indices (Figure 1) and combined into “classical” (i.e., SST, chl, and bathymetry), “mesoscale” (i.e., FSLE, OW, EKE, and SST and chl surface fronts), and “full” arrays (i.e., all features). After initial tests, k-means (kmeans, stats package, http://cran.r-project.org/; [42]) was determined to be the most robust cluster analysis algorithm to objectively classify biogeochemical subprovinces. This partitioning method, using Euclidean distances, assigns data points to k clusters and minimizes the sum of squares between the data points to cluster centre. With this algorithm, k must be defined a priori. In order to define k, we bootstrap (1000 times) k between 2 and 30. The between-clusters sum of squares is then divided by the total sum of squares to find the explained sum of squares. Arbitrary 1% and 5% thresholds are defined (Figure S1 in File S1), which we used to define the optimal k for the three multivariate arrays (Table 1), whereby the explained sum of squares for each additional k increases by less than 1% and 5%, respectively. K-means analyses were then performed on each array using the optimal k for both threshold levels (1%; Figure S2 in File S1 and 5%; Figure 2). The resultant clusters were defined as the biogeochemical subprovinces of the Mediterranean and Black Seas as a subdivision of the Mediterranean provinces defined by [33].


Defining Mediterranean and Black Sea biogeochemical subprovinces and synthetic ocean indicators using mesoscale oceanographic features.

Nieblas AE, Drushka K, Reygondeau G, Rossi V, Demarcq H, Dubroca L, Bonhommeau S - PLoS ONE (2014)

Biogeochemical subprovinces of the Mediterranean and Black Seas.Subprovinces for the (a) “classical”, (b) “mesoscale”, and (c) “full” multivariate arrays using a 5% threshold for the explained sum of squares to define the optimal number of subprovinces (see text).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4216069&req=5

pone-0111251-g002: Biogeochemical subprovinces of the Mediterranean and Black Seas.Subprovinces for the (a) “classical”, (b) “mesoscale”, and (c) “full” multivariate arrays using a 5% threshold for the explained sum of squares to define the optimal number of subprovinces (see text).
Mentions: Multivariate arrays were created from time-averages of the monthly scaled oceanographic indices (Figure 1) and combined into “classical” (i.e., SST, chl, and bathymetry), “mesoscale” (i.e., FSLE, OW, EKE, and SST and chl surface fronts), and “full” arrays (i.e., all features). After initial tests, k-means (kmeans, stats package, http://cran.r-project.org/; [42]) was determined to be the most robust cluster analysis algorithm to objectively classify biogeochemical subprovinces. This partitioning method, using Euclidean distances, assigns data points to k clusters and minimizes the sum of squares between the data points to cluster centre. With this algorithm, k must be defined a priori. In order to define k, we bootstrap (1000 times) k between 2 and 30. The between-clusters sum of squares is then divided by the total sum of squares to find the explained sum of squares. Arbitrary 1% and 5% thresholds are defined (Figure S1 in File S1), which we used to define the optimal k for the three multivariate arrays (Table 1), whereby the explained sum of squares for each additional k increases by less than 1% and 5%, respectively. K-means analyses were then performed on each array using the optimal k for both threshold levels (1%; Figure S2 in File S1 and 5%; Figure 2). The resultant clusters were defined as the biogeochemical subprovinces of the Mediterranean and Black Seas as a subdivision of the Mediterranean provinces defined by [33].

Bottom Line: Principal components analysis is then performed on the oceanographic variables to define integrative indices to monitor the environmental changes within each resultant subprovince at monthly resolutions.The first axis of the principal component analysis is explained primarily by classical ocean features and the second axis is explained by mesoscale features.Biogeochemical subprovinces identified by the present study can be useful within the European management framework as an objective geographical framework of the Mediterranean and Black Seas, and the synthetic ocean indicators developed here can be used to monitor variability and long-term change.

View Article: PubMed Central - PubMed

Affiliation: Unité Mixte Recherche Ecosystèmes Marins Exploités 212, Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER), Sète, France.

ABSTRACT
The Mediterranean and Black Seas are semi-enclosed basins characterized by high environmental variability and growing anthropogenic pressure. This has led to an increasing need for a bioregionalization of the oceanic environment at local and regional scales that can be used for managerial applications as a geographical reference. We aim to identify biogeochemical subprovinces within this domain, and develop synthetic indices of the key oceanographic dynamics of each subprovince to quantify baselines from which to assess variability and change. To do this, we compile a data set of 101 months (2002-2010) of a variety of both "classical" (i.e., sea surface temperature, surface chlorophyll-a, and bathymetry) and "mesoscale" (i.e., eddy kinetic energy, finite-size Lyapunov exponents, and surface frontal gradients) ocean features that we use to characterize the surface ocean variability. We employ a k-means clustering algorithm to objectively define biogeochemical subprovinces based on classical features, and, for the first time, on mesoscale features, and on a combination of both classical and mesoscale features. Principal components analysis is then performed on the oceanographic variables to define integrative indices to monitor the environmental changes within each resultant subprovince at monthly resolutions. Using both the classical and mesoscale features, we find five biogeochemical subprovinces for the Mediterranean and Black Seas. Interestingly, the use of mesoscale variables contributes highly in the delineation of the open ocean. The first axis of the principal component analysis is explained primarily by classical ocean features and the second axis is explained by mesoscale features. Biogeochemical subprovinces identified by the present study can be useful within the European management framework as an objective geographical framework of the Mediterranean and Black Seas, and the synthetic ocean indicators developed here can be used to monitor variability and long-term change.

Show MeSH