|
Affiliation: Massachusetts Institute of Technology, Bjerknes Centre for climate Research, U. Bergen
Event Date: Thursday, January 11, 2007
Location: NASA GSFC, Building 33, Room E125
Time: 1:00 - 2:00 PM
Inverting The Ice Sheet Problem: What Is The Ice Flow That Fits The Data?
Models and data are often used side by side to answer scientific questions, but rarely together. Ice sheet modeling provides a classic example, with surface data used as boundary conditions to dynamic ice models, but it is unclear whether the resulting flow field actually represents reality. Data assimilation attempts to use model and data together to constrain the flow and the initial model state to find a temperature and velocity field that best fits observations.
We review a case study in data assimilation from the global ocean. That experiment sought to minimize the difference between the model flow fields and the combination of WOCE section data, TOPEX/POSEIDON surface elevation data and surface forcing fields over the period 1992 - 1997. This was achieved by adjusting the initial model temperature and salinity and the time-dependent surface fluxes, while keeping those "control vectors" within the range of accepted uncertainties.
The ice sheet data assimilation project that we are proposing would use the dynamic ice sheet model SICOPOLIS for Greenland. This model uses the shallow-ice approximation, a surface degree-day model, Glen's flow law, a sliding law at the bedrock, heat conduction and isostatic adjustment in the lithosphere. A preliminary requirement to data assimilation is the derivation of the model's "adjoint" or derivative code. SICOPOLIS' adjoint was derived using TAF, a code-to-code conversion tool. This model adjoint has the ancillary benefit of providing an extremely efficient approach to calculating model sensitivities; for example the sensitivity of the ice volume to grid point by grid point surface temperatures or the sensitivity of the calving volume to the details of the bottom sliding parameterization. For the data assimilation experiment, the data would include surface GPS measurements, surface velocities from satellite radar interferometry, surface melt extent, and borehole data.
Posted or updated: Friday, January 5, 2007
Editor: Paul Przyborski
|
