HR: 0830h
AN: C11B-0816    [PDF]
TI: Paleothermometry via a Genetic Algorithm With an Application to GISP2 Borehole Data
AU: * Kim, Y
EM: ykim@geosci.uchicago.edu
AF: Department of Geophysical Sciences, University of Chicago, 5734 S. Ellis Ave, Chicago, IL 60637 United States
AB: Borehole temperature-depth profiles of ice sheets contain information about local climate change stretching back more than a hundred thousand years to the Eemian. Surface temperature histories at the Greenland ice sheet drill sites, GRIP and GISP2, have been reconstructed in the literature using a Control Method (Cuffey et al., 1995) and an Inverse Monte Carlo method (Dahl-Jensen, et al., 1998). In contrast to previous inversion efforts, the flexibility of the proposed Genetic Algorithm based method makes it possible to easily incorporate other climate proxies obtained from the ice core itself as constraints into the inversion process either by augmenting the fitness function with appropriate penalty terms or by treating it as a multi-objective optimization problem. The Genetic Algorithm based surface temperature inversion is applied to synthetic borehole temperature profiles and the GISP2 borehole data. Comparisons between the reconstructed and prescribed temperature forcing in the synthetic cases demonstrate the robustness of the new inversion method. Inversion of GISP2 data shows that the retrodicted 40,000 year long surface temperature history obtained by the Genetic Algorithm agrees well with results of previous studies while using only minimal problem specific knowledge. The resulting surface temperature reconstruction is used as an independent calibration of the linear oxygen-isotope paleothermometer ($\delta^{18}O = \alpha T + \beta$) and yields a value for $\alpha = 0.360$ per mil $\deg C^{-1}$ and $\beta = -23.9$ per mil. The Genetic Algorithm based inversion method enables a multi-proxy approach to ice sheet borehole temperature inversions which should improve the accuracy and resolution of surface temperature reconstructions and aid in the interpretation of very long $\delta^{18}O$ records from ice cores as a local temperature proxy.
DE: 1827 Glaciology (1863)
DE: 3260 Inverse theory
DE: 3344 Paleoclimatology
DE: 9315 Arctic region
SC: Cryosphere [C]
MN: 2003 Fall Meeting