Reconstruction of Holocene climate change using image analysis of laminated sediments
Concern about anthropogenic climate change has created an urgent need for preinstrumental climate data. Reliable instrumental data only extend back to c. AD 1850, so pre-instrumental proxy data are essential to allow the study of climate variability on decadal, centennial and millennial timescales. Laminated sediment cores can potentially supply annual-resolution palaeoclimate proxy data. Techniques have been developed to allow the application of high-resolution digital image analysis to annually-laminated Holocene sediment cores from two basins: Drammensfjord, Norway, and Lake St Moritz, Switzerland. Core chronologies have been produced by integrating digital-image-based varve counting, radiocarbon dating, and other independent chronological data. Image processing techniques have then been used to measure sediment greyvalue, which can be linked to sediment composition, and lamination thicknesses, which can be linked to changes in sediment accumulation. These data have been converted to evenly-sampled time-series, and have then been analysed using spectral analysis and comparisons with instrumental climate data in order to interpret the climatic influences that they record. The Drammensfjord record indicates that the onset of anoxia in Drammensfjord occurred at c. AD 1000. The thicknesses of the grey varve sublayers are a proxy for the magnitude of the annual spring flood. A significant increase in the frequency of large spring floods in c. AD 1470 may be linked to the onset of the Little Ice Age. There is no signal of the North Atlantic Oscillation in the Drammensfjord varves. Spectral analysis reveals "solar" periodicities in the varve data, but a solar explanation is unconvincing. The Lake St Moritz record indicates only a weak correlation between varve thickness and temperature. Sediment greyvalue is closely linked to sediment total organic carbon content, allowing a high-resolution proxy record to be produced. Spectral analysis of this indicates a red noise spectrum with possible significant periodicities at c. 2,870, 250, and 113 years, but interpretation of these is difficult because of the multiple climatic, anthropogenic and lake-ontogenetic influences on the record.