Development and testing of transfer functions for generating quantitative climatic estimates from Australian pollen data

Autor: Ellyn J. Cook, Sander van der Kaars
Přispěvatelé: Earth and Climate
Jazyk: angličtina
Rok vydání: 2006
Předmět:
Zdroj: Journal of Quaternary Science. John Wiley and Sons Ltd
Cook, E J & Kaars, S V D 2006, ' Development and testing of transfer functions for generating quantitative climatic estimates from Australian pollen data ', Journal of Quaternary Science . https://doi.org/10.1002/jqs.1076
ISSN: 0267-8179
DOI: 10.1002/jqs.1076
Popis: We review attempts to derive quantitative climatic estimates from Australian pollen data, including the climatic envelope, climatic indicator and modern analogue approaches, and outline the need to pursue alternatives for use as input to, or validation of, simulations by models of past, present and future climate patterns. To this end, we have constructed and tested modern pollen–climate transfer functions for mainland southeastern Australia and Tasmania using the existing southeastern Australian pollen database and for northern Australia using a new pollen database we are developing. After testing for statistical significance, 11 parameters were selected for mainland southeastern Australia, seven for Tasmania and six for northern Australia. The functions are based on weighted-averaging partial least squares regression and their predictive ability evaluated against modern observational climate data using leave-one-out cross-validation. Functions for summer, annual and winter rainfall and temperatures are most robust for southeastern Australia, while in Tasmania functions for minimum temperature of the coldest period, mean winter and mean annual temperature are the most reliable. In northern Australia, annual and summer rainfall and annual and summer moisture indexes are the strongest. The validation of all functions means all can be applied to Quaternary pollen records from these three areas with confidence. Copyright © 2006 John Wiley & Sons, Ltd.
Databáze: OpenAIRE
Externí odkaz: