Funds are provided to produce an eddy-permitting Arctic and sub-polar North Atlantic state estimate (ASTE) for climate research, using basic tools developed within the Estimating the Circulation and Climate of the Ocean (ECCO) consortium. The coupled ocean/sea-ice general circulation model (MITgcm) will be constrained by as many ocean and sea-ice observations as available and practical. The target period is from 1992 to present. The fit will be achieved through minimization of a least-squares misfit function (adjoint or Lagrange Multiplier method). The coupled ocean/sea-ice adjoint model has been generated by means of automatic differentiation (AD). A similar effort has been successfully accomplished with the production of the Southern Ocean State Estimate (SOSE). The domain of the ASTE covers the Atlantic northward of 26 N and the entire Arctic, with the exception of the Pacific sector southward of Bering Strait (66 N). The horizontal resolution is 7 to 12 km with 50 vertical levels. The control space comprises the initial conditions of the coupled ocean/sea-ice state, the time-varying surface atmospheric state, time-varying open boundary conditions, and spatially varying model parameters.

The observational backbone used initially for the ocean consists of: sea surface height from satellite altimetry, daily satellite SST, mean geoid and time-varying bottom pressure from satellite gravity and moorings, Argo float profiles, hydrographic data from WOCE and PHC, available mooring arrays such as Line-W and RAPID/MOCHA, and current meters available via AON/CADIS, PANGEAE (ASOF), and NSIDC (SCICEX). Sea-ice satellite observations used initially will comprise: daily ice concentration, ice drift velocities from passive microwave radiometry, scatterometry, and SAR imagery. In-situ under-ice data will include ice-tethered profilers, ice mass balance buoys, ice draft from submarine-borne upward looking sonar, as well as various buoy and mooring data from the IABP and AON/CADIS archives. Extension to other data sources includes tests of satellite retrievals of ice thickness or freeboard data from combined radar and laser altimetry.

Their approach offers strict dynamical consistency of the computed state estimate over the entire integration period, enabling exact time-varying budget calculations relevant for climate studies, such as the assessment of circulation shifts and their associated lateral heat and freshwater changes on seasonal and longer time scales. In this sense, the product differs significantly from what in the atmospheric context are called ?re-analysis? products. Another novelty of the product is its combined treatment of the Arctic and the subpolar gyre, and the synthesis of all available data types. Initial science foci will be on the connection between deep water formation processes and the Atlantic meridional overturning, the freshwater input at high latitudes and its pathways, and the interaction between Atlantic Water, Arctic halocline formation, and sea-ice.

Project Report

Introduction The Arctic and sub-polar ocean is home to several distinct processes and phenomena of importance to the global climate system. We have developed a regional eddy-permitting Arctic/subpolar gyre State Estimate (ASTE) using gradient-based optimization methods that are well developed in the computational science community, and that have been successfully applied for global-scale ocean climate state estimation by the Estimating the Circulation and Climate of the Ocean (ECCO) consortium. The approach is, fundamentally, a least-squares fit of allavailable observations to a coupled ocean/sea-ice model by means of the adjoint method (Heimbach and Wunsch 2012; Wunsch and Heimbach 2013b). The underlying model used is the MIT general circulation model (MITgcm). The fit of the model to the observations is achieved through variation of a number of control variables, such as initial conditions; time-mean parameter fields; and atmospheric forcing fields. Accomplishments 1. The project has contributed to the production of a global 1992-2011 state estimate on a new-generation global LLC90 grid. The grid includes the Arctic and provides a natural 3:1 nesting with the higher-resolution LLC270 grid. By way of example, Fig. 1 depicts the time-mean misfit between in-situ potential temperature observations and the state estimate at three depth levels for a non-optimized (a) and optimized (b) solution. 2. Fenty et al. (2013a,b) conducted studies of mass and enthalpy budgets and associated ocean-ice-atmosphere fluxes during wintertime sea ice quasi-equilibrium state in the Labrador Sea and Baffin Bay, a critical region of the North Atlantic subpolar gyre,, and analyzed what sets the maximum sea ice quasi-equilibrium extent (Fig. 2). 3. Nguyen et al. (2011,2012) inferred sources, pathways, and seasonal development of Upper Halocline Waters in the Western Arctic ocean from Pacific water intrusions via Bering Strait, and studied the implications for heat advection through the Chukchi Sea and Canada Basin. 4.. The potential use of first-year (FY) versus multi-year (MY) ice area retrievals from QuikSCAT as data constraints in ASTE requires an approach to simulate sea ice age and its classification according to FY and MY ice (Fig. 3, top panels). Such an algorithm has recently been incorporated into the MITgcm (Rampal et al., unpublished). Fig. 3 (bottom) shows apparent trends in satellite-retrieved (black line) and simulated (blue line) MY sea ice area fraction changes between 2000 and 2009. 5. A related study investigated the cause of IPCC AR4 models to underestimate thinning rates of sea ice thickness over the last three decades. To this end kinematic properties were intercompared, and compared with to SCICEX and ICESat thickness data as well as IABP buoy drift data. The causes of the model deficiency were identified to be a too weak feedback between the sea ice state and its kinematic properties. 6. With the provision of open boundary conditions to ASTE from the global solution, an understanding of basic transport properties becomes a priority. An initial such analysis was performed by Wunsch and Heimbach (2013a) for the Atlantic meridional overturning circulation (AMOC) covering the period. At 26N, where the RAPID/MOCHA array exists, the AMOC exhibits large month-to-month variability (Fig. 4, black curves), as pointed out by Wunsch and Heimbach (2006). The transport estimates have been contributed to the NOAA State of the Climate Report 2012 (Baringer et al., 2013). Note that in contrast to RAPID, the ASTE estimates enable a wide range of Atlantic basin scale analyses (Heimbach et al., 2011). They reach back to 1992, showing no discernible trend over the that period. 7. Buckley et al. (2013) analyzed low-frequency SST variabilityand upper-ocean heat content in the North Atlantic. The analysis showcases the quantitative rigor feasible through the availability of state estimates with accurately closed property budgets. Detailed contributions of heat content variability from local air-sea heat fluxes, Ekman transport, geostrophic transports, and diffusive processes are inferred for the sub-polar and sub-tropical gyres. 8. Chaudhury et al. (2013,2014) provided a comprehensive analyses of variability of atmospheric forcing variables. Detailed bias analyses from available reanalysis products and corresponding satellite data in the Arctic from 1992--2011 led to 50 percent reduction in sea ice thickness misfits when compared to ICESat derived thickness. 9. Recent attention on Greenland ice sheet-ocean interaction has focused on the mechanisms through which ocean forcing can impact glacier stability (Straneo et al. 2012,2013). Relatively little attention has been paid to the causes of recent warming of the North Atlantic. The only comparable warm period was in the 1930s -- when limited existing records indicate that similar widespread glacier retreat occurred. The present warming is due to a combination of a warm phase of the AMO, NAO, intrinsic ocean variability, and a long-term warming trend associated with accumulation of heat by the tropical ocean (Fig. 5; Straneo and Heimbach, 2013).

National Science Foundation (NSF)
Division of Polar Programs (PLR)
Standard Grant (Standard)
Application #
Program Officer
William J. Wiseman, Jr.
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Massachusetts Institute of Technology
United States
Zip Code