HR: 10:20h
AN: PP52A-01 INVITED     [Abstracts]
TI: Using the Paleorecord to Evaluate Climate-Model Performance in Projecting Changes in Climate Variability
AU: * Bartlein, P J
EM: bartlein@uoregon.edu
AF: Univ. Oregon, Dept. Geography, Eugene, OR 97403-1251 United States
AB: Changes in the variability of climate are likely to have greater societal impact over time than will simple changes in the mean state of climate, because that variability directly controls the frequency of droughts, floods, and other extreme events, and both directly and indirectly (through disturbance) affects terrestrial ecosystems. Syntheses of paleoclimatic data have been used in documenting and understanding changes in the mean state of climate, such as those that occurred during the Holocene in response to orbital forcing. Comparisons of climate-model simulations with the paleodata in that case provided critical information for the task of projecting climate changes by documenting the need to include feedbacks among coupled systems when simulating a climate different from the present day. There are two approaches for comparing paleoclimatic simulations with paleodata. The inverse approach, in which paleodata are interpreted in climatic terms, has been most frequently applied for inferring past changes in the mean state of climate. There are several problems with this approach, including the observations that the distributions of biotic indicators and the amplitude of geochemical or sedimentological indicators are usually governed by climatic extremes, and that most indicators have multiple proximal controls, which are usually interrelated in a nonlinear fashion. Chronological issues make it difficult to synchronize individual records and therefore limit our ability to detect changes in teleconnection patterns or climate modes that are defined by spatial patterns. The alternative forward approach applies offline or coupled environmental submodels to produce model output that is directly comparable with paleodata. Ideally, a global network of temporally synchronized, annual- (or subannual-) resolution records would be available for comparison with model simulations, but such a network is unlikely to be developed soon. In the meantime, time-slice syntheses or individual non-synchonized records should be the targets for data-model comparisons. With the employment of appropriate tools and experimental designs, it may be possible to use time-slice data syntheses to examine the ability of models to correctly simulate changes in climate variability in addition to changes in the mean state.
UR: http://geography.uoregon.edu/envchange/
DE: 3344 Paleoclimatology
DE: 1851 Plant ecology
DE: 1610 Atmosphere (0315, 0325)
DE: 1615 Biogeochemical processes (4805)
DE: 0400 Biogeosciences
SC: Paleoceanography and Paleoclimatology [PP]
MN: 2004 AGU Fall Meeting