Glacial and Holocene climates of Australia reconstructed by vegetation-model inversion
Abstract
We present reconstructions of temperature and moisture variables for Australia at key periods in the last glacial and the Holocene. The reconstructions were made by inversion of a simple, semi-empirical plant-functional type (PFT) based vegetation model developed using a Process-Oriented Niche Specification (PONS) approach, which makes it possible to take into account the effect of changing orbital parameters and CO2 concentration on plant water relations and therefore on the climatic implications of observed vegetation shifts. The data are PFT abundance ';scores' derived from site-based pollen assemblages in nearly 2000 records. The procedure used to calculate these scores involves assigning all living taxa that could contribute to a given pollen taxon to PFTs (independently, based on major plant traits), and using Bayes' theorem to re-allocate the share of the pollen sum represented by ';ambiguous' taxa to the different PFTs they could represent, given information on the other taxa present. In each case the vegetation model was run with a large set of alternative climates corresponding to systematic perturbations around the modern climate. The reconstructed anomaly of each climate variable is the difference between the best-fit climate (for which the dissimilarity between the modelled and observed profile of PFT scores is a minimum) and the modern climate at the site. This minimum is identified by fitting a second-degree polynomial surface to the dissimilarity values as a function of climate variables. The method also allows the assignment of a confidence interval using the second derivative of the fitted surface. We demonstrate that this inversion technique reproduces modern climates from surface samples collected at the fossil pollen sites, with only modest uncertainties, implying that the reconstructions are plausible. This work provides the first quantitative reconstructions of climate changes across Australia over the last glacial-interglacial cycle, and provides evaluation targets for palaeoclimate-model simulations that have been conducted as part of the Coupled Modelling Intercomparison Project (CMIP5). The method is potentially of more general application, as it combines some of the desirable features of model inversion approaches (above all, the ability to incorporate effects of [CO2] on vegetation) with the conceptual simplicity and closeness to observations of statistical calibration or response-surface methods. Moreover, it looks forward to ';next-generation' vegetation models in which plant distributions and abundances will be characterized in terms of functional traits.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFMPP54A..03H
- Keywords:
-
- 1637 GLOBAL CHANGE Regional climate change;
- 3344 ATMOSPHERIC PROCESSES Paleoclimatology;
- 0473 BIOGEOSCIENCES Paleoclimatology and paleoceanography