d’upwelling de Humboldt par l’activité intra-
saisonnière
V.1.Préambule
Dans les chapitres précédents, nous nous sommes intéressés à des mécanismes d’interaction d’échelles impliquant l’ensemble du système couplé du Pacifique tropical. En particulier, nous montrons d’une part que la propension du système couplé à produire des événements extrêmes (qui est liée à la nature même d’ENSO) dépend de l’état moyen, et d’autre part que les nonlinéarités d’ENSO (mesurées par cette même propension du système couplé à produire des événements extrêmes) peuvent induire un résidu participant à la modulation basse fréquence de l’état moyen. Ce mécanisme est transposé au cycle saisonnier de température et à sa modulation : en particulier nous montrons que les cycles saisonniers de la variance d’ENSO et de ses nonlinéarités sont modulés à basse fréquence. Nous nous intéressons ici à un autre type d’interaction d’échelles, celle entre la variabilité intra- saisonnière océanique et l’état moyen et la circulation moyenne dans le Pacifique tropical est. Cette étude est motivée par deux aspects : 1) d’une part, parmi l’ensemble des processus étudiés jusqu’à présent dans cette thèse, nous souhaitons étendre le spectre des échelles de temps vers les haute fréquences pour l’étude des processus nonlinéaires dans les tropiques, 2) d’autre part, la région du Pacifique tropical est au cœur des préoccupations de la communauté scientifique qui s’intéresse à la variabilité climatique (cf. projet VOCALS/VAMOS/CLIVAR). En particulier, cette région pourrait être le siège de processus dit d’upwscalling par lesquels la variabilité à l’échelle régionale peut influencer la variabilité climatique (cf. Toniazzo, 2010). Elle accueille par ailleurs l’écosystème le plus riche du point de vue des ressources halieutiques (cf. Figure V.1), ce qui en fait une région sensible du point de vue des impacts sociétaux.
V.1.1.Eastern tropical Pacific climatic specificities
The Eastern Equatorial Pacific (hereinafter EEP) is characterized by a zonal band of equatorial minimum Sea Surface Temperature (SST) known as the Cold Tongue (hereinafter CT). The latter appears as an outstanding feature of the EEP as it can be considered as an extension of the cold upwelled waters of the Humboldt Current System (HCS hereinafter), located off Peru and Chile coasts. The CT region is of significant interest for oceanographic and climatic studies. On the first hand, as mentioned by many recent works (Fiedler and Talley, 2006; Karnauskas et al., 2007, among others), the CT plays a significant role in global hydrological and biogeochemical cycles, impacting the formation of cloud, precipitation and thus the surface and near-surface oceanic productivity in this region. These topics are currently under investigation within the frame of the VOCALS international program2. On the other hand, it can be considered as a key element connecting the equatorial Pacific variability to the HCS climate variability. Actually, the South American coast behaves as an extension of the equatorial wave-guide on a wide range of frequencies (Shaffer et al., 1997; Pizarro et al., 2001, 2002; Vega et al., 2003). The HCS, which represents the most productive Eastern Boundary Currents region (Figure V.1), is then directly under the influence of equatorial perturbations. The mechanisms by which remote equatorial forcing impacts the regional circulation of the coast of Peru and Chile are not yet fully understood by the scientific community.
2
The VAMOS Ocean-Cloud-Atmosphere-Land Study (VOCALS) is an international CLIVAR program the major goal of which is to develop and promote scientific activities leading to improved understanding of the South Eastern Pacific coupled ocean-atmosphere-land system on diurnal to inter-annual timescales.
Figure V.1. Primary production in Chlorophyll-a concentratin (mg/m3). Mean over 1995-2005. Data are from the SEAWifs satellite.
In particular, most Ocean General Circulation Models (OGCMs) and Coupled General Circulation Models (CGCMs) produce a significant cold bias in this CT region, along with an exaggerated westward extent (Stockdale et al., 1998; Meehl et al., 2001; Karnauskas et al., 2007; Ye et Hsieh, 2008). Not surprisingly, this cold bias is also reported in the tropical part of the HCS, i.e. in the Peru System (Mechoso et al., 1995; Maes et al., 1997). While some authors suggest the importance of lateral and vertical mixing (Maes et al., 1997; Cravatte et al., 2007, Noh et al., 2005) in modifying the structure of equatorial currents and thus the dynamics in this region, others put forward the relevance of the bathymetric feature represented by the Galapagos Islands (GI hereinafter) (Eden and Timmerman, 2004; ET04 hereinafter; Karnauskas et al., 2007).
Actually, since there are directly located on the equator, this archipelago provides a topographic barrier for Southern Equatorial Current (SEC) and Equatorial Under Current (EUC) (see Strub et al. (1998) for a quick view of these currents pathways). Because the thermocline is shallow in this region, a realistic representation of the mean equatorial currents and of the vertical stratification is required for properly simulating SST (Clement et al., 1996). Moreover, it has been demonstrated in recent studies based on lagrangian diagnostics that water masses which compose the EUC, feed in part the Peru Chile Under Current (PCUC) (Croquette, 2006; Montes et al., 2010). This also emphasizes the strong link between
the equatorial variability and the HCS variability. Despite the possibility that the equatorial circulation may be changed by topographic features such as the GI, the latter hardly reach the surface in most of OGCMs/CGCMs, mainly due to the too coarse longitudinal resolution. This has the potential to lead to a limited skill of these models to represent the CT vertical structure and thus its dynamics. Indeed, by obstructing the EUC, GI constrains the equatorial Pacific thermodynamical adjustment in a way that tends to reduce the meridional overturning circulation, and therefore the entertainment mixing, leading to a warmer SST in the CT region (Karnauskas et al., 2007; hereinafter K07). Considering the sensitivity of the tropical Pacific system to the zonal contrast of SST along the equator (in particular, it determines the ENSO stability, cf. Jin and An, 1999), these processes at regional scales have to be considered/evaluated with regards to their rectified effect on the mean state.
High resolution regional modelling is required to tackle such issues. Actually, an increased resolution will allow to properly include GI bathymetry and to enhance mesoscale processes involved in lateral/vertical mixing and diffusion and thus to better account for the CT dynamics.
V.2.Methodology
V.2.1.Regional modelling with ROMS
The numerical quasi-equilibrium solutions we analyzed were obtained with the regional model ROMS (Regional Oceanic Modeling System) (Shchepetkin et McWilliams, 2003). The primitive equations are discretized on a three dimensional Arakawa (C type) grid with horizontal curvilinear and vertical sigma coordinates. Sigma coordinates are terrain- following to allow a better vertical resolution than z-coordinates. However, these coordinates lead to some discrepancies in the computation of horizontal pressure gradient in the vicinity of strong slopes. This issue has recently been addressed by splitting advection scheme from diffusion scheme, the latter being represented by a 4th order biharmonic scheme (cf. Marchesiello et al., 2009). Barotropic and baroclinic components are also spitted, with a smaller time-step for the latter component. The vertical mixing is parameterized with a KPP profile (KProfile Parametrization, cf. Penven et al., 2006). The vertical diffusion terms are solved through a semi-implicit Crank-Nicholson scheme. The 3rd order upstream biased