In order to provide usable data to (paleo-) climate modellers, important efforts must be performed to obtain high-resolution quantified observations from geologic sequences. Until now, description of sedimentary structures are commonly qualitative and vague. Therefore, a method using image analysis has been developed and tested to quantitatively characterize sedimentary microstructures.
First, a freeze-drying technique to cut thin-section without disturbance of soft-sediments structures is elaborated.
Analysis of microscopic images has not yet been used widely in the Earth Sciences. Indeed, difficulties in applying the technique arise mainly because of the nature of the images to be analysed. For instance, it is commonly difficult to distinguish boundaries between adjacent colourless grains, between small superposed features of interest, or between textural features and blurry cement or sedimentary matrix. Moreover, visual observation alone is commonly insufficient to identify obscure textural features. Techniques for image analysis must therefore be chosen carefully and adapted to the optical characteristics of each specific sample collection. Those difficulties are minimized in this new technique.
Optical and backscattered electron microscope photographs are digitised from thin-sections. Processing of the 256 grey-scale pictures produce binary (black and white) images, where white pixels represent the clay-rich sedimentary matrix and black pixels represent objects contained within this matrix: silt- or sand- sized terrigenous or neomorphosed particles, diatoms or organic debris. From objects selected by the operator, several parameters are measured: gravity centre, surface, axes length, both of best-fitting ellipse, angle between long axis of ellipse and horizontal, and perimeter.
From these measurements, several indices are defined in order to quantify certain aspects of the objects forming the sedimentary structures as for instance their size, shape, orientation or fabric. The quantified observations obtained using image analysis compare well with the descriptions made by conventional methods.
Potentials of this technique are illustrated through the study of sediments from three lakes.
At Lake Baikal, micro-bioturbations can be detected in thin-sections. A bioturbation quantification is proposed: a bioturbation indice, named H, is defined. To provide a calibration, H indice is applied on conventional laminated and bioturbed sedimentary fabric.
Other structures attributed to turbidites or sub-lacustrine "levee" can also be identified and quantified via this technique. For instance, quantified grain-size measurements are made at the lamina scale and produce objective high resolution observations in shorter time than conventional methods.
At Vico (Latium, Italy), micro-bioturbation is also detected and quantified. Image analysis quantification reveals a probable influence of the hypolymnion oxygenation conditions on magnetic minerals neomorphism.
At Nogaret, (Massif Central, France), an automated counting of the laminations and their thickness is successfully used. Records are seasonal and allow reconstruction of annual climatic variation, and more particularly reveal series of 'terribly wet summer'.