Characterising lake sediments by NIRS: integrating teaching and research in the Universities Central Teaching Laboratory

A trend of ‘brownification’ identified in surface waters across the UK, North America and Northern Europe has been a growing concern, with the transfer of organic carbon from catchments to lake basins and water courses increasing. To explore this problem, in 2012-3 graduate teaching assistant Fiona Russell began PhD research investigating whether the sedimentary records preserved in lake basins provided new information on the longer term onset of this trend, the nature of any previous events in the last 11,500 years and the causes of browner surface waters. This research has used lakes in the UK Uplands Water Monitoring Network (formerly known as Acid Waters Monitoring Network), who for 30 years have collected data that shows some of these acid sensitive sites are affected by increases in the flux of dissolved organic carbon (DOC) and browner waters. The research involved accessing the sediments from these lakes and exploring for patterns that matched recent increases in monitored DOC and for other changes in the organic stratigraphy that might explain the patterns.

Humic waters on the edge of Loch Grannoch

 At this time, the award winning state-of-the-art Central Teaching Laboratories (CTL) at the University of Liverpool expanded the Infrared Spectrometry capabilities adding a Near-infrared Diffuse Reflectance machine. The Bruker MPA Diffuse Reflectance Fourier Transform Near Infrared Spectrometer (FT-NIRS) was added to the teaching facility to enable the rapid and non-destructive characterisation of powders including soils and sediments. NIRS has proven to be a valuable tool for quantifying the components present in lake sediments, such as type and quantity of organic and mineral matter. Often the approach uses direct comparison of the NIR spectra to parallel independently measured parameters, for example organic content (loss-on-ignition or total organic carbon). This approach requires the development of an extensive training set of parallel measurements, and one has been developed in the CTL to enable undergraduates and postgraduates to estimate loss-on-ignition by NIRS. The NIRS measurement takes 60 seconds and is non-destructive, whereas equivalent combustion methods take several hours, are destructive and more difficult to accommodate in the time frame of practical classes.

MPA FT-NIRS in the Central Teaching Laboratory

In developing utilities to support teaching in the School of Environmental Sciences, Fiona Russell began exploring alternative approaches to the handling of NIR spectra to obtain a broader range of environmental information from sediments and soils. The approach developed rather than using a large training set, for example our loss-on-ignition dataset comprises > 200 samples with parallel NIRS and l-o-i measurements, instead uses a library of end member NIR spectra measured for known composition materials e.g. biogenic silica (diatoms), humic/fulvic acids (key components of DOC), rock and mineral types. This alternative approach, recently published in the Journal of Paleolimnology (click here), applies a multiple regression of a selection of known material NIR spectra to a sample set of NIR spectra measured for unknown composition materials, in this research lake sediment samples. The research shows that with appropriate selection of 3-5 end member spectra, the method can reconstruct successfully the proportions that these end members form in the series of unknown lake sediments. This different approach means that environmental reconstructions can be undertaken using 3-5 end member spectra rather than a library of >100-200 parallel measurements and it can reconstruct the proportions of multiple end member materials or sediment components simultaneously.

This slideshow requires JavaScript.

To demonstrate the approach, we sampled a 14,500 year duration sediment core from Loch Grannoch in the Galloway Hills (SW Scotland). Loch Grannoch is a small upland oligotrophic lake that has granite bedrock, relatively large catchment area and is one of the lakes in the UK Uplands Water Monitoring Network. Applying our approach reveals a down-core pattern of varying dominance by biogenic silica, organic and mineral content from the late glacial to present. Testing a wider range of end members shows there is sensitivity to the choice of end members, with a local bedrock or sediment end member characteristic of the lake catchment important. The wider generality of the approach, and the utility of our growing library of mineral, organic and biogenic silica end-member materials, is demonstrated by successful ‘blind’ application to additional lake sediment cores from Wales, Norway and Sweden each with differing climate and bedrock.

Fitting mineral, organic and biogenic silica end members to multiple lakes.

The technique development embedded in this research showcases the importance of integrating teaching and research at the University of Liverpool. The paper is available as Open Access from the Journal of Paleolimnology should you wish to read more about it. The team have made available the library of end member NIR spectra and the multiple regression method code developed using the open source R platform with the the online version of the paper supplementary materials and in the University of Liverpool DataCat archive. The methodology is now being used routinely by colleagues, undergraduate and postgraduate students to characterise materials in their learning and independent research. For further information please contact the authors: Fiona Russell, John Boyle and Richard Chiverrell.