 |
| Titel |
Reconstructing temperatures from lake sediments in northern Europe: what do the biological proxies really tell us? |
| VerfasserIn |
Laura Cunningham, Naomi Holmes, Christian Bigler, Anna Dadal, Jonas Bergman, Lars Eriksson, Stephen Brooks, Pete Langdon, Chris Caseldine |
| Konferenz |
EGU General Assembly 2010
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 12 (2010) |
| Datensatznummer |
250038610
|
|
|
|
|
|
| Zusammenfassung |
| Over the past two decades considerable effort has been devoted to quantitatively
reconstructing temperatures from biological proxies preserved in lake sediments, via
transfer functions. Such transfer functions typically consist of modern sediment
samples, collected over a broad environmental gradient. Correlations between the
biological communities and environmental parameters observed over these broad
gradients are assumed to be equally valid temporally. The predictive ability of such
spatially based transfer functions has traditionally been assessed by comparisons of
measured and inferred temperatures within the calibration sets, with little validation
against historical data. Although statistical techniques such as bootstrapping may
improve error estimation, this approach remains partly a circular argument. This raises
the question of how reliable such reconstructions are for inferring past changes in
temperature?
In order to address this question, we used transfer functions to reconstruct July
temperatures from diatoms and chironomids from several locations across northern Europe.
The transfer functions used showed good internal calibration statistics (r2 = 0.66 - 0.91). The
diatom and chironomid inferred July air temperatures were compared to local observational
records. As the sediment records were non-annual, all data were first smoothed using a 15 yr
moving average filter. None of the five biologically-inferred temperature records were
correlated with the local meteorological records. Furthermore, diatom inferred temperatures
did not agree with chironomid inferred temperatures from the same cores from the same
sites.
In an attempt to understand this poor performance the biological proxy data was
compressed using principal component analysis (PCA), and the PCA axes compared
to the local meteorological data. These analyses clearly demonstrated that July
temperatures were not correlated with the biological data at these locations. Some
correlations were observed between the biological proxies and autumn and spring
temperatures, although this varied slightly between sites and proxies. For example,
chironomid data from Iceland was most strongly correlated with temperatures in
February, March and April whilst in northern Sweden, the chironomid data was most
strongly correlated with temperatures in March, April and May. It is suggested that the
biological data at these sites may be responding to changes in the length of the ice-free
period or hydrological regimes (including snow melt), rather than temperature per
se.
Our findings demonstrate the need to validate inferred temperatures against local
meteorological data. Where such validation cannot be undertaken, inferred temperature
reconstructions should be treated cautiously. |
|
|
|
|
|
|