The biogeochemical model
The biogeochemical Eco3M-S model is a multi-nutrient and multi-plankton functional type model (Auger et al., 2011). It describes the cycles of C, N, P and Si. This model was previously used to study the multi-annual ecosystem variability in the NW Med (Auger et al. 2014; Herrmann et al. 2013) and the ecosystem functioning in the Gulf of Lions shelf influenced by the Rhone river inputs. We have re-calibrated the previous version described in Auger et al. (2014) in order to implement the model on the whole Mediterranean Sea. The different recalibrated parameters are described in (Kessouri et al., in prep-b). Eco3M-S consists of seven interconnected compartments (Fig. 2): (1) The inorganic compartment representing nitrate, ammonium, phosphate and silicate, (2) the dissolved organic matter compartment considered under the forms of carbon, nitrogen and phosphorus (3) particulate organic matter compartment (under the forms of C, N, P, Si and chlorophyll), divided in two weight classes, light and heavy, (4) the autotrophic compartment composed of three classes of phytoplankton classified by size: pico-phytoplankton (smaller than 2 microns, nano-phytoplankton (between 2 and 20 microns), and micro-phytoplankton (between 20 and 200 microns) that corresponds in the model to diatoms, (5) the zooplankton compartment composed of three size classes of zooplankton: nano-zooplankton [diameter < 20µm], micro-zooplankton [20µm > diameter > 200µm] and meso-zooplankton [diameter> 200µm] and (6) the bacteria compartment. The relative internal composition, i.e. the stoichiometry, of each functional type is considered as variable for autotrophic organisms and constant for heterotrophic organisms. (7) Oxygen: Oxygen is consumed by plankton and bacteria; it is also produced inside the water mass through phytoplankton photosynthesis. Air-sea exchanges are performed throughout Wanninkhof and McGillis (1999) laws.
Biological boundary conditions
In the Atlantic Ocean, nutrients and oxygen have been prescribed using monthly profiles from “World Ocean Atlas 2009” climatology at 7.5° W. This boundary condition’s limit has been chosen close to the Strait of Gibraltar to avoid spurious effects caused by relaxation on physical parameters in the buffer zone.
The Black Sea was not included in the model domain but it represents a major source of freshwater. It was considered in the NEMO-MED12 configuration as a river for the Aegean Sea at the Dardanelles Strait. At this strait, constant values for organic matter have been applied according to Copin Montegut (1993). Nutrient inputs have not been introduced at the Dardanelles Strait because of the null net flow (Tugrul et al., 2002).
At the runoff points, concentrations of nutrients have been imposed by region using dataset of Ludwig et al. (2010). As organic matter concentrations in the continental waters are most often unknown, constant values derived from the historical dataset of the Rhône River at Arles station (MOOSE program, pers. comm. Raimbault) have been imposed for these inputs.
At the sea surface, atmospheric deposition of inorganic matter has been prescribed as constant values for western (west to Sicily Strait) and eastern sub-basins based on “low” estimations reported by Ribera d’Alcalà et al. (2003). Deposition of organic matter has been deduced from relationships between inorganic and organic matter deduced from MOOSE network data collected south of France (pers. comm. Raimbault).
Nutrient fluxes at the water column/sediment interface have been obtained through a coupling of the biogeochemical model with a simplified version of the vertically-integrated benthic model described by Soetaert et al. (2000).
Initialization and spin up
In previous Mediterranean Sea modelling studies, initialization states were based either on climatology (Crise et al., 1998; Crispi et al., 2002) or, at sub-basin scale, on relationships between nutrients and physical properties of water masses (Prieur and Legendre, 1988; Auger et al., 2014). In this study, inorganic variables have been initialized using the historical database defined in 13 regions by Lavezza et al. (2011). Depending on the amount and spatial distribution of nutrients and hydrological salinity profiles in each bio-region, we have chosen one type of initialization mentioned above: we have used relations linking nutrients to salinity in the upper layer and intermediate waters, then a constant value in the deeper layer in the western basin where profiles are numerous and allow a seasonal description. In the eastern basin, because of the lack of the combined biological and physical datasets near coastal zones we have deduced nutrients’ median profiles in each region defined by Lavezza et al. (2011).
The spin-up of a model is an important issue as its stability is required to calculate reliable budgets. Stable nutrient and organic matter stocks on the whole the basin have been obtained after four years of simulation (for instance, the total nitrate stock presents an annual variability around 0.05 %). Thus, the simulation has been started in 2000 using the final state of a first 2000-2004 simulation. It ends in 2013. In this paper we focus on the period 2003 – 2013.
This section is divided in four parts:
First, we present an evaluation (1) of simulated surface chlorophyll patterns using daily 4km resolution outputs of MODIS Aqua satellite products and (2) of water mass properties, focusing on nutrient chlorophyll and DOC concentrations over the basin, based on the trans Mediterranean BOUM cruise dataset (Pujo-Pay et al., 2011; Moutin and Prieur, 2012) occurring in summer 2008. To address the lack of temporal comparison, especially in winter, values were reported and compared to experiences mentioned in literature in the Gulf of Lions, in the Adriatic Sea and in the Rhodes Gyre.
In the second part, the monthly climatologies of hydrological and biogeochemical parameters calculated over the [2003 – 2013] decade from model outputs are presented and discussed. In the third part, the functioning of four kinds of regimes is described throughout clustering analysis using models output using selected parameters. Finally, matter exchanges and flux at the straits of the Mediterranean Sea and biological fluxes in the eastern and the western basins are highlighted.
Biogeochemical characteristics of water masses
The ability of the model to represent the biogeochemical characteristics of the different water masses has been evaluated trough a confrontation of model results against the observations of the BOUM cruise (Moutin & Prieur, 2012) along transects from North to South in the western basin and from West to East in the eastern basin (indicated on Figure 1). The cruise lasted for one month and a half between June and July 2008. We have averaged the model outputs during the period of the cruise (July to August 2008).
Over the entire transect, observed and simulated nutrients (nitrate and phosphate) show extremely low concentrations in the surface layer (Fig. 5). Nitracline has been defined here as the depth where nitrate is equal to 1 mmol.m-3, and for the phosphacline the threshold concentration has been chosen equal to 0.05 mmol.m-3 (Lazzari et al., 2012). As in the observations, nitracline is localized at 60 m depth in the western basin and gradually deepens to attain 120 m in the eastern Levantine basin. Phosphacline is close to the nitracline in the western basin but a gradual decoupling between both nutriclines is then observed to the east. The difference between both depths could reach 40 m in some locations in the eastern part of the Levantine Sea.
Net primary production (NPP)
The NPP is heterogeneous over the Mediterranean Sea (Fig. 10). Highest values are estimated in the Alboran Sea, where PPN never decreases under 0.5 gC.m-2.d-1 and attains locally 1 gC.m-2.d-1 in winter and in spring. At the same time, the WAG contains very lower production which is caused by its anticyclonic circulation. This anticyclone is surrounded by the most productive dynamics of the Mediterranean Sea in a frontal zone. Minima are estimated in the extreme eastern Levantine basin by 0.1 gC.m-2.d-1 in autumn and 0.4 gC.m-2.d– in winter. The NPP is particularly heterogeneous in the NW, where northern gyre (NG) and the surrounding area show antagonist patterns in winter and spring. In winter, low PPN is simulated in the NG while high values are visible in peripheral zones, and inversely in spring. When the high production triggers, the surrounded waters become less productive. The NPP reaches inside the NG 0.2 gC.m-2.d-1 in winter and 0.9 gC.m-2.d-1 in spring. These values are close to estimates from observations described by Uitz et al. (2012).
NPP seasonal evolution is different from the one of the surface chlorophyll concentrations: high production is simulated at the end of spring and beginning of summer (Fig. 10) when the lowest values are simulated and observed for surface chlorophyll (Fig. 4). In the Algerian basin, there is accumulated nutrients in the euphotic layer. During this period, chlorophyll profiles get globally a homogeneous form but some local features such as eddies favor “high surface chlorophyll” (HSC) profiles shape described by (Lavigne et al., 2015) when nutriclines become shallow (not shown).
Main pelagic ecological regimes of the Mediterranean Sea
In the previous section, climatologies of hydrodynamic and biogeochemical variables have been presented in order to gain in understanding of the functioning of the pelagic ecosystem of the Mediterranean Sea. Large spatial heterogeneities have been noticed in both north-south and west-east directions. Here we propose to use a statistical method of clustering to simplify this complex picture and to identify bio-regions characterized by similar seasonal cycles. This objective identification of bioregions is a step forward compared to a fixed division of regions based on geography. Such a regionalization has been proposed for the Mediterranean by D’Ortenzio and Ribera D’Alcalà (2009) from a detailed analysis of the surface chlorophyll seasonal cycle. The different seasonal cycles found with this classification identified different “trophic regimes” present in the classification of Longhurst (1998). They also showed that the structure of the seasonal cycle is tightly coupled with the dynamic range of biomass characterizing the productivity of the region. The idea of repeating this exercise with a model is to have benefit of the large volume of information given by a coupled model not only at the surface but also in the sub-surface as the values of the surface chlorophyll are negligible during a large part of the year although primary production is still active in the DCM.
For clustering, we classified each pixels on the basis of the climatological values of the surface and depth-integrated chlorophyll, the primary production, the nitrate uptake, the amount of nitrate in the upper layer [0-150m], nitrate concentration in the intermediate layer (500 m), the mixed layer depth and the stratification index. Then, for each bioregions we calculated the mean time series. The first three parameters inform about the biological surface and subsurface properties, the nitrate uptake about the ability of phytoplankton to consume nutrients, the amount of nitrate represents the balance between the nutrients imports by vertical dynamics and their consumption and the last parameter characterize the global patterns of the hydrology.
No-bloom regimes of the eastern basin (groups 4, 5, 6)
The eastern basin includes the No-bloom northeastern, southeastern and Ionian bioregions (Fig. 12). It is characterized by deep DCM in summer and weak nitrate supply to the surface. Deep nutrient concentrations are 40% less important than in the western basin.
In winter, phytoplankton blooming depends on phosphacline depth pierced during mixing events as mentioned by (Lazzari et al., 2012). As shown in Fig. 14, MLD exceeds the phosphacline depths over short periods in February and March for groups 4 and 5. In the no-bloom southeastern regime (group 6), the averaged MLD is never located under the top of the phosphacline. These regimes display low primary production at the beginning of winter. The model overestimates it with values generally in the range 0.45 – 0.55 gC.m-2.d-1 between January and March while (Uitz et al., 2012) gave estimates between 0.15 and 0.45. Like in the western no-bloom regime, vertical mixing allows a weak enrichment by nutrients to maintain subsurface phytoplankton production. The new production represents 40 to 50 % of the total production at this period.
During the stratified period, in summer, the DCM reaches its maximum depth. The DCM depth is increasing from west to east. Maximum depth (120 m) is found in the Levantine Sea in August and September (Figs. 8 & 14). Chlorophyll concentration in the DCM is low and decreases from spring to summer from 0.4 to 0.25 mg m-3 (Fig. 8). The regenerated production represents about 80% of the total production.
In the no-bloom regimes of the Eastern basin, phosphacline gradually deepens related to the nitracline towards the east. The maximum difference between the phosphacline and nitracline can reach 48 m in the Levantine Sea in February and March. This sub-basin presents the poorest surface and subsurface waters of the Mediterranean in terms of nutrients concentrations and biological activity.
The Intermittent/Intermediate regime (group 2)
This regime characterizes waters surrounding the deep convection area of the northwest Mediterranean as well as regions with moderate to strong mixing (Bonifacio gyre, Adriatic) and stratified regions of the Alboran Sea. The first category is characterized by permanent currents associated to density fronts, as the Northern Current, the North Balearic front where winter mixing can reach 200 m. In the north Alboran Sea, the MLD is shallow all over the year but the vertical dynamics promoting nutrient enrichment is strong at the periphery of the permanent mesoscale features. Such a variety of regions in this cluster has been also highlighted by DR09.
The common point to these regions is probably an enrichment which is intermediate between the bloom and the no-bloom regimes. This is clear on Fig. 14 where the MLD is significantly deeper than the nutriclines during winter while, first, in the no-bloom regions these interfaces are closer from each other, and, second, in the bloom region, the difference is much larger. Another consequence is that the MLD is generally not deep enough to inhibit production (the southern Adriatic should be an exception). Whatever the injection of nutrients is done by convection (Adriatic) or by frontal dynamics (NW and north frontal zone of the Alboran Sea) with immediate consumption. In the first case, the MLD is significantly deeper than in the second one, and in the second case, the frontal dynamics allows to inject nutrients directly in the MLD.
However, it can be noted an important difference between the two types of regions forming this group. In the “mixing” regions like the Adriatic, the primary production is apparently lower than in the frontal regions (see Alboran) but a larger part of this production is attributed to new production (maximum f-ratio of 0.66 in February in the Adriatic and only 0.39 in January in Alboran).
Table of contents :
1. Table des matières
1 CHAPITRE 1 : Introduction générale
1.1. La Méditerranée – Généralités
1.2. Objectifs de la thèse
1.3. Structure du manuscrit
2 CHAPITRE 2 : Description hydrodynamique
2.1. Description géographique des divers bassins
2.2. Structure verticale des masses d’eau
2.3. Circulation océanique en Méditerranée
2.4. La convection profonde
3 CHAPITRE 3 : Interaction entre dynamique et biogénèse
3.1. Les sels nutritifs
3.1.1 Les échanges des sels nutritifs
3.1.2 Impact du mélange sur les échanges verticaux de sels nutritifs
3.1.3 Enrichissement de surface par la convection profonde
3.2. La couche de mélange et le cycle de la chlorophylle
4 CHAPITRE 4 : Méthodologie
4.1. Etat de l’art sur la modélisation couplée physique biogéochimie en Mer Méditerranée
4.2. Les modèles couplés physique-biogéochimie
4.2.1 Les champs physiques
4.2.2 Le modèle biogéochimique (Eco3m-S)
4.2.3 Le couplage physique-biogéochimie
4.3. Les grilles de calcul
4.3.1 La grille à l’échelle du bassin méditerranéen
4.3.2 La grille du bassin occidental
5 CHAPITRE 5 : Résultats
1. Introduction Biogeochemical cycles of the Mediterranean Sea: climatologies and budget
2.1. The physical model
2.2. The biogeochemical model
2.3. Implementation on the Mediterranean basin
2.3.1. Biological boundary conditions
2.3.2. Initialization and spin up
3. Results and discussion
3.1. Evaluation of the model
3.1.1. Surface chlorophyll
3.1.2. Biogeochemical characteristics of water masses
3.2.1 Mixed layer depth
3.2.2 Depth and magnitude of the Deep Chlorophyll Maximum (DCM)
3.2.4. Net primary production (NPP)
3.2.5. Organic carbon export
3.3. Main pelagic ecological regimes of the Mediterranean Sea
3.3.1. Bloom like regime (group 1)
3.3.2. No-bloom regime of the western basin (group 3)
3.3.3. No-bloom regimes of the eastern basin (groups 4, 5, 6)
3.3.4. The Intermittent/Intermediate regime (group 2)
3.4. Nitrogen and phosphorus dynamics in the Mediterranean Sea
3.4.1. Biogeochemical processes
3.4.2. Fluxes at the straits
3.4.3. The global budget
4. Conclusion and perspectives
1. Introduction : Nitrogen and phosphorus cycles in the western Mediterranean Sea using high resolution modeling: Processes and budget
2.2.3. Initial and boundary conditions
2.2.4. Derived variables
3. Results and discussion
3.1. Evaluation of the simulation
3.1.1. Surface chlorophyll
3.1.2. Horizontal and vertical biogeochemical patterns in winter and spring
3.1.3. Variability between April and September 2013
3.1.4. Annual cycle of the nutrients stocks
3.2. Atmospheric forcing and hydrology
3.3. Consequences of hydrological variability on biogeochemical processes and stoichiometry
3.4. Budget of the nitrogen and phosphorus in the northwestern Mediterranean basin
Appendix – Model parameters
2. Introduction Phytoplankton dynamics and biogeochemical fluxes in the western Mediterranean Sea using a 3D physical/biogeochemical coupled model
3.1. The hydrodynamic model
3.2. The biogeochemical model
3.3. Areas of study
3.4. MODIS satellite dataset
3.5. Determination of the bloom onset
3.6. Calculation of the F-ratio
4. Model evaluation
5. Results and discussion
5.1. Atmospheric conditions
5.2. Vertical mixing and chlorophyll concentration
4.2.1. The Northern Gyre
4.2.2. The shallow convection area
4.2.3. The stratified region
5.3. Conditions at the onsets of the phytoplankton efflorescence
5.4. Primary production
5.5. Particulate organic carbon export
4.6 The trophic regimes
8. Bibliographie supplémentaire
6.1 Annexe 1 : Déroulement de la thèse
6.2 Annexe 2 : Les bases de données et la calibration
2.1. Base de données Lavezza et al (2011) pour l’initialisation des sels nutritifs, et initialisation de
2.2. Le produit satellite MODIS
2.3. Les flotteurs BioArgo
2.4. Rappel de l’utilisation des bases de données pour la calibration et perfectionnement des
6.3 Annexes 3 : Les « groupes de travail (WP)» du chantier Méditerranéen MISTRALS et la
6.4 Annexes 4 : Paramètres du modèle biogéochimique