transfer functions in palaeoclimatology

Home > Earth and the Environment > Atmosphere and Weather > Weather and Climate: Terms and Concepts

transfer functions in palaeoclimatology Transfer functions are frequently employed in Quaternary palaeoclimatology because they provide an important method of reconstructing past climatic conditions. Transfer function equations are derived by using multivariate statistical techniques to define the relationship between biological or chemical entities and a physical parameter (which is usually an environmentally sensitive variable such as temperature) from within a large and diverse modern-day calibration data set. The application of these equations to geological palaeodata enables a quantitative estimate of the physical parameter to be calculated for the past. Transfer functions have been constructed for numerous faunal and floral biological groups, including beetles, coccoliths, foraminifera, molluscs, and pollen. Transfer functions therefore facilitate the investigation of biosphere–climate interactions in both the terrestrial and marine realms and support palaeoclimatic reconstruction on a global scale.

The underlying rationale behind the transfer function approach is that biological organisms are preferentially adapted to a particular set of ecological and environmental conditions and that the distribution and composition of species assemblages can therefore be related to certain variables within the physical environment. The successful application of a transfer function is, however, also dependent on certain assumptions being satisfied:(i) the distribution of the biota has to be closely linked to the predicted environmental variable; (ii) the modern calibration data set must provide an accurate representation of the species–climate relationship; and (iii) the ecological requirements of the various taxa being studied cannot vary appreciably in the time interval under consideration.

Transfer functions can be calculated from qualitative, semi-quantitative, or quantitative data, although the greater objectivity and precision associated with quantitative data make them the most suitable for multivariate statistical analysis. Similarly, variations in the composition of species assemblages, rather than the distribution record of a single taxon, are preferable since they can better account for the complexity of factors that may affect the ecological response of an individual species and hence improve the precision of the transfer function.

The Imbrie–Kipp transfer function approach is a commonly applied method; this technique was used by members of the CLIMAP project to calculate past variations in ocean surface temperature using data for planktonic microfossil assemblages (i.e. data for, foraminifera, coccolithophorida, and Radiolaria) from deep-sea cores. The transfer function is calculated in two steps: (i) a factor analysis of individual species abundance data from the core top calibration data set to generate ecological groupings or assemblages; and (ii) a regression analysis of the assemblage factor loadings with modern sea surface temperature to produce a set of temperature equations which can subsequently be applied to fossil assemblages.

A number of alternative methods for calculating transfer functions also exist. For example, the modern analogue technique employs similarity coefficients (e.g. squared Euclidean distance) to evaluate the differences between the palaeoproxy data and the modern calibration dataset. The palaeoenvironmental estimate is calculated as a simple or a weighted average of the climatic values from the most similar modern subset. Transfer functions derived using the mutual climatic range method, which are also determined using similarity coefficients, differ in that they are based on qualitative presence or absence of data.

Mark R. Chapman

palaeoclimatology, transfer functions in see transfer functions in palaeoclimatology