Hyperspectral, discrete and XRD analyses on Adamello glacier ice-core sections focusing on dust layers

Ice-covered surfaces of the planet are a pristine glimpse on the climate history of the Earth. Their capability to reflect the solar radiation affects the energy budget through time, so that paleoclimatic information recorded in polar and mountain glaciers tell a story about climate and environmental changes. Climate variations are a glaring reality due to a progressive climate warming that has been observed since the 1950s. The latest 2022 IPCC report [1], highlights the worrying state of the entire cryosphere health. Mid-latitude glaciers, e.g., are highly sensitive to the current atmospheric warming, which is seriously compromising the quality of the signal preserved in the ice. Mineral dust aerosol in atmosphere can alter the Earth’s energy budget interacting with the solar electromagnetic radiation. The dust settled on glaciers surfaces varies the ice capability to reflect the incident radiation, so that a major portion of it is absorbed as a function of Light Absorbing Impurity (LAI) concentration, mineralogical composition, LAI shapes etc. Depending on formation sites different minerals have different optical properties, affecting cryosphere mechanisms in a way that has to be further investigated. Ice core studies from mid-latitude mountain glaciers are essential to infer recent climate variability and anthropogenic impact on a regional scale. In the scientific literature, various imaging non-destructive systems extract e.g. information from visual stratigraphy of ice cores [2]. Advanced techniques can further be developed to improve the accuracy of ice cores measurements, pursuing the preservation of such precious achieves. The aim of this work is to apply different approaches to improve the understanding and climate simulation models regarding ice optical properties and feedback mechanisms, focusing on mineral dust and ice internal features. First, a non destructive system based on a hyperspectral imaging sensor is used to analyze optical properties of ice-core sections at the EuroCold Laboratory (University Milano-Bicocca). Hyperspectral imaging spectroscopy is a powerful technique used to characterize material’s surfaces on the basis of their optical properties. Considering that different features vary the ice capability to reflect the electromagnetic radiation (e.g. the visible light), in this context three parameters will be extracted from reflectance: Albedo (the amount of energy reflected by the ice), Snow Darkening Index (SDI) and Impurity Index (II) both used to characterize dust. The SDI record will be useful to select samples for discrete dust concentration analyses and X-Ray Diffraction (XRD) measurements. Finally, a calibrated curve for concentration will be created. We consider ice-core sections, extracted by the project ClimADA in 2021 from the Adamello-Mandrone glacier group region, for a length of about 59.8 m and a depth ranging from 3.4 m to 63.2 m. The hyperspectral set-up is a scanning system made by different components: movable and fixed ones [3]. The main mobile parts are a hyperspectral spectrometer (Hyperspec® VNIR, HeadWall Photonics), which collects spectral radiance in 840 bands in VIS and NIR wavelengths ranging from 380 to 1000 nm, and a halogen stable light source (600 W, LOT Quantum Design). Both the spectrometer and the source move back and forth to scan ice-core sections at a defined speed, appositely chosen to provide the right micro-step motion pursuing high spatial resolution images. The White Reference (a white Spectralon panel), positioned at the top of each section on the fixed flat system’s surface, reproduces an ideal reflectance equal to 1 (maximum value) to be associated with a Dark Current measure that reproduces a 0 reflectance (minimum value). Trough data acquisition and processing software, ice structures and LAI can be characterized using the three different spectral descriptors (Albedo, SDI and II) derived from reflectance spectrum for each image’s pixel of a reflectance matrix [4]. Record of the SDI index, as mentioned above, is also used to select samples for discrete and XRD analyses. For discrete data, a Coulter Counter Beckman Multisizer 4E works with particle sizes from 1 𝜇𝑚 to 30 𝜇𝑚. Here 4 groups of SDI peaks are considered: high SDI peaks, medium-high SDI peaks, high albedo peaks and medium-high albedo peaks. These peaks are respectively associated to high impurity content, medium-high impurity content, high ice light reflection and presence of air bubbles. Finally, the XRD instrument works with an incident X-Ray tube beam directed toward LAI, which diffracts the beam with an angular distribution depending on the ordered crystalline lattice that forms the material itself. A detector moves around the system to acquire the angular directions and intensities of the outgoing diffracted ray. This method will be applied on highly concentrated dust layers. Averaging on horizontal bands of pixels moving along the core, it’s possible to see how reflectance and optical descriptors vary depending on ice features (Fig.1). Dust strongly decreases the albedo and ice lenses rapidly increase it, while SDI has the opposite link with dust. Bubbles have a particular behavior, halfway between dust and lenses. The section considered here is of particular interest due to the high concentration of impurity contained. Thanks to Coulter Counter measurements, for Fig.1 the estimated concentration is (12.93 ± 3.67) 𝑝𝑝𝑚 for the dust portion, (0.58 ± 0.40) 𝑝𝑝𝑚 for lenses and (0.19 ± 0.09) 𝑝𝑝𝑚 for parts with air bubbles. The lower concentration when bubbles are present, if compared to lenses, might seems in contrast with the reflectance curves trend. However, bubbles affect ice optical properties in few ways that have yet to be further clarify due to different bubbles dimensions, regularity, contained gasses, bubbles number etc. At this point, a regression model is applied and the calibrated curve is extracted (Fig.2), to highlight the forecasted regions with low, medium and high impurity concentration for many sections of the ice-core. The concentration ranges from 0 ppm to a maximum of about 30 ppm. The XRD analysis, in addition, gives in output a spectrum with each peak associated to precise minerals. Using the “search-match” qualitative approach, it’s possible to identify all peaks above the background (Fig. 3). The shape of peaks is associated to various characteristics, such as minerals concentration, lattice internal defects and minerals orientation. From peaks, we can extract the relative concentrations of minerals; here we have 35.6% Biotite, 22.8% Kaolinite, 20.8% Clinochlore, 10.9% Quartz and 9.9% Albite. The hyperspectral method allows a dust spectral characterization in a non-destructive way. Highly concentrated dust layers produce an SDI increase, facilitating the samples selection process for discrete and mineralogical analyses. XRD is a powerful tool to extract data for the identification of the main mineral groups within a sample. This can give clues regarding dust source regions to track back where the dust comes from, improving the understanding of atmospheric patterns and thus enhancing the atmospheric transport simulation models. Furthermore, the unification of Hyperspectral, discrete and XRD analyses could allow the identification of reflectance curves that work as footprints related to precise mineral groups, to create a reference for additional studies. However, the incident electromagnetic radiation penetrates for a few millimeters within the surface creating multiple scattering emissions, which could generate reflectance artifacts such as abrupt indices variations that have to be considered. Nevertheless, instrumental improvements could reduce all the unwanted effects. The integration of hyperspectral imaging analyses with classic ice core measurements will allow a comprehensive view of chemical and physical properties of ice cores all over the world.