Near-infrared spectrometry (NIRS): A new tool for inferring past climatic changes from lake sediments

Abstract

This study tests the hypothesis that lake sediments contain climate-related information that can be detected by near-infrared spectrometry (NIRS), and that NIRS can be used to infer past climatic changes from analysis of sediment cores. NIRS is a rapid and non-destructive technique that measures attributes of the chemical composition of organic materials. A training set of 76 lakes from northern Sweden, spanning a broad altitudinal gradient, was used to assess whether lake altitude and vegetation zones can be modelled from NIR spectra of surface sediments (0-1 cm) using partial least squares (PLS) regression and soft independent modelling of class analogies (SIMCA) classification. Lake altitude served as a surrogate variable reflecting differences in climatic conditions among sites. After spectral filtering using orthogonal signal correction (OSC), cross-validated predictions explained 86% of the variance in altitude and the prediction error (root mean square error) was 78 m, corresponding to 8.3% of the gradient (390-1250 m above sea level). To evaluate the significance of NIR spectral differences between surface sediments of lakes in different vegetation zones (mountain-birch forest, dwarf shrub and alpine heath), principal component analysis (PCA) models were developed separately for lakes in each vegetation zone. Multivariate classification analysis demonstrated that NIR spectra of surficial sediments differed between lakes located in different vegetation zones. A separate sediment data set from 56 lakes was used to assess sediment ageing effects on NIR signals. Marked similarities between NIR spectra in surface sediments (0-1 cm) and sediments from 1-2 cm depth indicated that degradation of organic material following sediment consolidation resulted in little loss or change of climate-related information. Finally, to assess the ability of NIRS methods to reconstruct past climatic changes over Holocene timescales, we applied the NIRS-altitude model to sediments in a core from a small mountain lake. Estimates of mean July air temperature based on the NIRS-altitude transfer function showed similar trends compared with inferences from chironomids, diatoms and pollen from the same core. Overall, the results indicate that changes in NIR spectra from lake sediments reflect differences in climate, and that NIRS models based on surface-sediment samples can be applied to sediment cores for retrospective analysis.