The Computational Research in Ice and Ocean Systems (CRIOS) group applies advanced mathematical methods and simulation techniques to improve our understanding of the role of the ocean, sea ice and polar ice sheets in the global climate system. The group’s research rests on the recognition that these Earth system components remain insufficiently sampled by observations. So the group works toward optimal extraction of information contained in available observations through formal synthesis with known dynamics. The dynamics, in turn, is encapsulated in the governing equations of motion and rendered in state-of-the-art numerical models. Improved mechanistic understanding along with uncertainty characterization are essential prerequisites for developing robust climate predictions with quantified uncertainties. CRIOS is involved in a number of formal and informal, national and international research collaborations, including the UK, Norway, Germany, Japan, and Australia. The group is pursuing several main research thrusts.
A central focus of CRIOS is on sustaining and improving the production of ocean state estimates to support global and regional ocean circulation and climate variability research on time scales of days to decades. CRIOS is a member of the NASA-funded "Estimating the Circulation and Climate of the Ocean" (ECCO) consortium, together with colleagues at JPL/Caltech, MIT, and AER. ECCO is one of the nation's leading global ocean data assimilation efforts. It has developed an advanced inverse modeling framework for synthesizing much of the available oceanographic satellite and in-situ observations of diverse types into a coherent framework of a state-of-the-art ocean general circulation model. The goal is to produce a best-possible description of the time-evolving global ocean circulation covering the era of satellite altimetry from 1992 onward. Some emerging research problems tackled by CRIOS members are:
(1) characterizing and understanding the dynamics and variability of the global and Atlantic Meridional Overturning Circulation (AMOC) in terms of the underlying three-dimensional ocean circulation;
(2) understanding the dynamics of global and regional sea level change, the role played by ocean dynamics in explaining regional sea level variability, and the partition of observed changes into thermosteric and mass contributions;
(3) developing a computational framework for characterizing and quantifying uncertainties associated with the sparse observations, the initial and boundary conditions, and the model parameters;
(4) developing formal methods for quantitative observing system design in support of regional and global observing campaigns.
A second broad area of research targets the marine-based or marine-terminating cryosphere. Through NSF funding we are improving simulations of coupled sea ice-ocean dynamics in the Arctic Ocean. Fundamental research questions here pertain to potential links between ocean circulation changes in the Arctic and North Atlantic and the decline of Arctic sea ice cover. A related research question is with regard to implied Arctic/subpolar North Atlantic freshwater balances and their role in subpolar North Atlantic ocean dynamics. We are also getting involved with coupling the ocean's physical state with biogeochemical, biological, and ecosystem properties.
A third area we are investigating is the polar ice sheets, their dynamics, their interaction with the ocean, and their contributions to sea level change. Of particular interest is understanding the role of glacier-fjord (Greenland) and ice shelf/stream-ocean interactions (Antarctica) in the observed mass loss of the Greenland and Antarctic ice sheets, and the potential for future abrupt destabilization and accelerated sea level rise. Another research thrust pertains to the development of dynamically consistent ice sheet (or coupled ice sheet-ocean) initialization schemes to enable skillful predictions.
Much of our work involves the use of advanced methods of computational science and engineering, in particular, formal inverse methods, state and parameter estimation, and uncertainty quantification. Some of these methods have in common the use of derivative forms of Earth system models, in particular adjoint and Hessian model codes. We are expanding the use of algorithmic differentiation-enabled adjoint code generation and maintenance for Earth system models.