Hyperspectral imaging spectroscopy: a promising method for the biogeochemical analysis of lake sediments

Abstract

Abstract. We investigate the potential of hyperspectral imaging spectrometry for the analysis of fresh sediment cores. A sediment-core-scanning system equipped with a camera working in the visual to near-infrared range (400 to 1000 nm) is described and a general methodology for processing and calibrating spectral data from sediments is proposed. We present an application from organic sediments of Lake Jaczno, a freshwater lake with biochemical varves in northern Poland. The sedimentary pigment bacteriopheophytin a (BPhe a) is diagnostic for anoxia in lakes and, therefore, an important ecological indicator. Calibration of the spectral data (BPhe a absorption ∼800 to 900 nm) to absolute BPhe a concentrations, as measured by high-performance-liquid-chromatography, reveals that sedimentary BPhe a concentrations can be estimated from spectral data with a model uncertainty of ∼10%. Based on this calibration model, we use the hyperspectral data from the sediment core to produce high-resolution intensity maps and time series of relative BPhe a concentrations (∼10 to 20 data points per year, pixel resolution 70 × 70 μm 2 ). We conclude that hyperspectral imaging is a very cost-and time-efficient method for the analysis of lake sediments and provides insight into the spatiotemporal structures of biogeochemical species at a degree of detail that is not possible with wet chemical analyses. © 2015 Society of Photo-Optical Instrumentation Engineers (SPIE)

A high-resolution pigment and productivity record from the varved Ponte Tresa basin (Lake Lugano, Switzerland) since 1919: insight from an approach that combines hyperspectral imaging and high-performance liquid chromatography

“…Following the approach of Butz et al (2015), the half cores were scanned with the Specim Ltd. single core scanner equipped with a visual to near infrared range (VNIR, 400-1000 nm) hyperspectral linescan camera (Specim PFD-CL-65-V10E). We normalized the raw scans with a BaSO4 white reference plate and extracted spectral endmembers according to the method of Kruse et al (1999) by using the spectral hourglass wizard of the software ENVI 5.03 (Exelisvis ENVI, Boulder, Colorado).…”

“…Based on the spectral endmembers, we calculated the relative absorption band depths (RABD) for an absorption feature between the reflectance at the wavelengths R590 and R730 (590-730 nm). In contrast to Butz et al (2015), who used a fixed band for the trough minimum, we used here the local minimum of the absorption feature for each individual spectrum (i.e. each pixel individually, spatial resolution 68 µm x 68 µm).…”

“…It is essential to ground-truth and verify remotely sensed data with independent established and quantitative analytical methods to properly interpret spectral indices (Butz et al 2015). Our case study from the Ponte Tresa sediments shows that a calibration between HSI-index data (here: RABD590-730), and absolute 'green pigment' concentrations {chlorophyll a + pheophytin a} measured with HPLC, reveals remarkable calibration statistics (R2 = 0.82; RMSEP ~ 12%) and that the methodology developed by Butz et al (2015) is applicable elsewhere in a different limnogeological setting.…”

“…Recently, important advancements have been made with non-destructive scanning reflectance spectroscopic techniques which facilitated the measurement of lutein, chlorophylls and their derivates (chlorins) directly from fresh sediment cores at mm-scale resolution (Rein and Sirocko 2002;Rein et al 2005;von Gunten et al 2009;Saunders et al 2012;Amann et al 2014). Butz et al (2015) and Butz et al (2017) developed this approach further and used spectra from a scanning hyperspectral imaging system (HSI; pixel size: ~68 x 68 µm; spectral resolutions ~2.7 nm) to infer 'green pigments' {chlorophyll a and pheophytin a} as well as bacteriopheophytin a at subvarve-scale resolution. The advantage of hyperspectral data is the very high spatial resolution; the disadvantage is that the speciation of the pigments is limited.…”

Tracking climate signals in varved lake sediments: research strategy and key sites for comprehensive process studies in the Masurian Lakeland

“…The impregnated slabs will be scanned using the micro-XRF technique (ITRAX Core Scanner) at the highest possible spatial resolution of measurements to obtain a sub-seasonal record of chemical element variability. VIS-RS scanning using a hyperspectral core scanner will provide an estimation of the variability in photopigments and clay mineral contents in the sediments (Butz et al 2015). In the next step, thin sections will be prepared from impregnate slabs and investigated for varve composition using a petrographic microscope.…”