3.1 Numerical Modeling Model Description We use the ORCA2_LIM3_PISCES configuration of the NEMO (Nucleus for European Modelling of the Ocean), a global ocean modeling framework which is composed of an ocean model, ORCA2, coupled to an ice model, LIM3, and a biogeochemichal model, PISCES. (Madec et al., 1999; Madec, 2008; http://www.nemo-ocean.eu/About-NEMO). Ocean model: ORCA2 is a primitive equation model adapted to regional and global ocean circulation problems. The distribution of variables is a three-dimensional Arakawa C-type grid. In the ORCA2_LIM3_PISCES configuration, the model domain extends from 78S to 90N. The model uses a global tripolar orthogonal curvilinear grid with 2 deg zonal resolution, and a meridional resolution varying from 0.5 at the equator to 2 cos(phi) south of 20S. The horizontal grid features two points of convergence in the Northern Hemisphere, both situated on continents to overcome the North Pole singularity found for geographical meshes, thus avoids singularity point inside the computational domain [Madec and Imbard, 1996], and attains the ratio of anisotropy nearly one everywhere. Minimum resolution in high latitudes is about 65 km in the Arctic and 50 km in the Antarctic [Timmermann et al, 2005]. In the vertical direction, the model uses a full or partial step z-coordinate, or s-coordinate. There are 31 levels, with 10 levels in the top 100m and partial steps in the lowest level. The vertical mesh is deduced from a mathematical function of z [Madec and Imbard, 1996]. Bottom topography and coastlines are derived from the ETOPO5 dataset and Smith and Sandwell (1997). In the ORCA2_LIM3_PISCES configuration, lateral tracer mixing is done along isopycnals. Eddy induced tracer advection is parameterized following Gent and McWilliams (1990). Horizontal momentum is mixed along model level surfaces using the eddy viscosity coefficients varying with latitude, longitude and depth. Vertical eddy diffusivity and viscosity coefficients are calculated using a 1.5 order turbulent kinetic energy model [Gaspar et al., 1990]. Zero fluxes of heat and salt and no-slip conditions are applied through lateral solid boundaries. At the bottom boundary, zero fluxes of heat and salt are applied through the ocean bottom. The ocean is subject to the external forcings of heat, freshwater, and momentum fluxes from the atmosphere and/or the sea ice. The solar radiation penetrates the top meters of the ocean. In the ORCA2_LIM3_PISCES configuration, the downward irradiance is formulated with two extinction coefficients [Paulson and Simpson, 1977], whose values correspond to a Type I water, the most transparent water, in Jerlov's classification [Jerlov, 1968]. We intend to adopt the parameterization proposed by Morel and Antoine [1994] to improve the upper ocean solar heating rate affected by the abundance of phytoplankton [Nakamoto, et al, 2001; Ueyoshi, et al, 2003; Ueyoshi, et al, 2005; Manizza, et al, 2005; Frouin, et al, 2007; Subrahmanyam, et al, 2007, Lengaigne, et al, 2007; Lengaigne, et al, 2009]. Solar radiation absorption and local heating within the upper layers of the ocean are strongly influenced by the abundance of phytoplankton. Depending on the degree of the chlorophyll concentration, the heat deposition occurs within an upper layer that may vary in thickness from less than 10 m to more than 100 m. Morel and Antoine's [1994] parameterization allows the vertical profiles of heating rate to be predicted from the phytoplanktonic pigment concentration. For the infrared waveband, not influenced by biological materials, the irradiance profile is described by a single exponential function. For the ultraviolet and visible (< 750 nm) band, a bimodal exponential form is adopted. The weights associated with each of these exponential functions, along with their specific attenuation lenghts, are dependent upon pigment concentration. These dependencies are explicated through polynomial formulas (Morel and Antoine, 1994). With the use of Morel and Antoine's parameterization the the upper ocean solar heating rate becomes temporally and spatically variable depending on pigment concentration. Sea-ice model: Within NEMO, the ocean is interfaced with the interactive sea-ice model LIM3 (Louvain-la-Neuve sea ice model) [Fichefet and Morales-Maqueda, 1997, Timmermann, et al, 2005]. LIM3 is a C-grid dynamic–thermodynamic model, including the representation of the subgrid-scale distributions of five-category sea ice thickness, enthalpy, salinity and age. Brine entrapment and drainage as well as brine impact on ice thermodynamics and a snow ice formation scheme are explicitly included (Vancoppenolle, et al, 2009a, Vancoppenolle, et al, 2009b). Biogeochemical model: The NEMO modeling framework includes PISCES based on the Hamburg Ocean Carbon Cycle 5 (HAMOCC5) model [Aumont et al., 2003; Aumont and Bopp, 2006] (see http://www.meece.eu/documents/deliverables/WP2/D2.4.pdf). PISCES is a biogeochemical model which simulates the marine biological productivity that describes the biogeochemical cycles of carbon and of the main nutrients (P, N, Si, Fe). PISCES assumes a constant Redfield ratio and phytoplankton growth is directly limited by the external availability in nutrients [Monod, 1942] so that the elemental ratios of Fe, Si and Chlorophyll (Chl) are prognostically predicted based on external concentrations of the limiting nutrients. PISCES has currently 24 compartments. These are five modeled limiting nutrients for phytoplankton growth: Nitrate and Ammonium, Phosphate, Silicate and Iron. Phosphate and Nitrate+Ammonium are not really independent nutrients in PISCES since they are linked by constant Redfield ratios but the nitrogen pool undergoes nitrogen fixation and denitrification. Four living compartments are represented: two phytoplankton size-classes/groups (nanophytplankton and diatoms) and two zooplankton size classes (microzooplankton and mesozooplankton). For phytoplankton, prognostic variables are total biomass, the iron, chlorophyll and silicon contents. This means that the Fe/C, Chl/C and Si/C ratios of both phytoplankton groups are fully predicted by the model. For zooplankton, only the total biomass is modeled. For all species, the O/C/N/P ratios are kept constant and are not allowed to vary. 3.2 Simulations of Climate Change Impacts during the last Decade The experiments using an Earth System (AOGCM) model with PISCES suggest that chlorophyll biomass strongly impacts the climate of the Arctic Ocean and improves the model sea-ice distribution (Lengaigne, et al., 2009). This impact results from direct biological heating through changes of turbidity in the vertical column of water within the upper layers of the ocean amplified by an indirect contribution involving enhanced ice melting. The phytoplankton blooms that occur concurrently with the ice retreat along the coastal shelves of the Arctic Ocean act in fact to trap the penetrating solar heat flux in the ocean surface layer, resulting in the warming of the SST along these coastal shelves. A reduction of sea-ice concentration in these regions triggered by this warming will increase the solar energy penetrating into the ocean, providing a positive feedback that further amplifies the direct biological heating. The ocean response is maximum in late summer, resulting in a warming of ~0.5 C along the continental shelves of the Arctic Ocean and reduction of the overall Arctic Ocean sea ice thickness (24%) and concentration (8%) (Lengaigne, et al., 2009). Similar results are reported in the experiments performed using an OGCM model with PISCES that indicate that phytoplankton biomas amplifies the seasonal cycle of SST, mixed-layer depth and sea-ice cover (Manizza et al, 2005). The bio-induced changes in the physical state such as a freshening of the Arctic Ocean due to increased melting, precipitation and river runoff and a slowdown of the overturning circulation are in turn likely to affect the biological processes (growth, grazing, remineralization). We'll therefore investigate the influence of these changes on the phytoplankton growth and concentration by performing the experiments described below. Furthermore, we'll attempt to account for these bio-physical feedbacks in greenhouse gases emmisions scenario experiments as described in Section 3.4 which should allow us to assess their impact on the Arctic sea-ice decline in the 21st century climate. EXP 1: Pigment values predicted by the PISCES model will be used for the heat rate calculation with Morel and Antoine's (1994) parameterization. EXP 2: Oserved sea ice concentrations (Comiso, 2007) and pigment values predicted by the PISCES model will be used with Morel and Antoine's (1994) parameterization. Each experiment will be run using two different atmospheric forcing datasets described below: (1): with the DFS4 dataset for 1958-2007. (2): with NCEP/NCAR reanalysis dataset for 1958-present (2010). Atmospheric forcing fields: The DFS4 dataset, a new set of 49-year-long (1958-2006) coarse-resolution (2 deg x 2 deg) forcing variables for OGCMs based on ERA40 will be used as atmospheric forcing fields at the ocean surface. The dataset has been extensively tested and proved that the refined ERA40 fields contributed to improve the representation of central features of the global ocean circulation [Brodeau, et al, 2010]. The dataset consists of 8 variables used for interannual forcing: 6-hourly wind velocity components at 10m, 6-hourly potential temperature and specific humidity at 2m, daily solar and infra-red downwelling radiation, and monthly total precipitation and snow. The dataset is currently extended to year 2007 and is available from the Drakkar group [The DRAKKAR Group, 2007; Brodeau, et al, 2006]. We also intend to use the CORE.2 Global Air-Sea Flux dataset and various climatologies available from NCAR along with NCEP/NCAR daily reanalysis data [Large and Yeager, 2004; (http://dss.ucar.edu/datasets/ds260.2/)]. Sea-ice response from the NEMO-LIM3 sea-ice model is found sensitive to these two different atmospheric forcings (Massonnet, et. al., 2010). We'll examine what different effects additionally these forcings will have on the biogeochemical model variables. We'll also perform an experiment (EXP 2) using the observed sea ice concentrations (Comiso, 2007) in place of the ORCA2_LIM3_PISCES model simulated ice concentrations and compare the differences in biogeochemical variables along with dynamical variables that results from the use of the modeled and observed sea ice concentrations. Biogeochemical data: We'll use various initial biogeochemical datasets described in Aumont and Bopp (2006). Biogeochemical datasets includes parameters such as Alkaline (Alk), Dissolved Inorganic Carbon (DIC), Nutrients (NO3, PO4, Si, Fer, NH4), Dissolved Organic Carbon (DOC) and O2. Spinup: One pass through the 1958-2007 forcing with the DFS4.1 dataset will be used as a spinup for the simulations beginning at 01Jan1958. In this “spin-up” run, the ocean is initially at rest with temperature and salinity set to January climatorogical values of Levitus (1982). In the Arctic Ocean, initial mean sea ice thickness of 3.5 m is imposed in regions with a sea surface temperature below 0 C. An initial ice concentration of 0.95 are also assumed (Vancoppenolle, et al, 2009). Initial biogeochemical parameters are also set following Aumont and Bopp (2006). From this initial state, the default ORCA2_LIM3_PISCES configuration of NEMO is integrated for 50 years with Newtonian restoring terms for sea surface salinity only. For the ocean surface forcing, the CORE bulk formula is used with the 50-year DFS4.1 dataset. A model experiment (Spinup) using the ORCA2_LIM3_PISCES has been conducted over the period 1958–2007 using the DFS4.1 ocean surface forcing datasets as described above. The model results for the period 1978-2007 of this experiment are used to produce the figures included in this proposal. Model output: The output variables from ORCA2 are usual temperature, salinity, etc. The PIECES model output variables include, among others, concentrations of 4 plankton functional types (nanophytoplankton, diatoms, meso and micro zooplankton) on the 31 model levels at various time frequencies from daily to yearly. Also included are pCO2, pO2, and alkalinity. The LIM3 variables include ice concentration, ice thickness, ice surface temperature, and snow thickness.