Figure - available from: The Cryosphere
This content is subject to copyright.
Comparison between nominal and measured concentrations of FD066 dust samples using the FlowCam and the Coulter counter. An orthogonal distance regression on the FlowCam data (black line with 3σ confidence interval in gray) shows good linearity over 3 orders of magnitude. The red line refers to the linear fit on the CC data. The y error bars reflect 1 standard deviation of multiple repetitions of the same sample. All x errors are estimated as 10 % of the FD066 prepared concentrations and account for the uncertainties in the dilutions and plastic adsorption effects. Both insets refer to the FlowCam measurements. The top inset shows the relative standard deviation (RSD) distribution, and the bottom inset shows the distribution of the residuals, defined as the difference between the expected and measured concentrations. The top bars indicate the approximate ranges of dust concentration in polar and mid-latitude records.
Source publication
Insoluble particles in ice cores record signatures of past climate parameters like vegetation dynamics, volcanic activity, and aridity. For some of them, the analytical detection relies on intensive bench microscopy investigation and requires dedicated sample preparation steps. Both are laborious, require in-depth knowledge, and often restrict samp...
Citations
... However, most methodologies failed for fossil pollen assemblages from lake sediments due to the large floristic diversity, fragmented, folded, and clumped grains, and the issue of discriminating microfossils of interest from other debris in fossil samples (e.g., amorphous organic matter, algal or invertebrate remains; Holt et al., 2011;Holt and Bennett, 2014;Khanzhina et al., 2018). Therefore, applications to fossil pollen samples, such as those from lake sediments (Bourel et al., 2020;Milošević et al., 2020;Theuerkauf et al., 2023) or ice cores (Maffezzoli et al., 2023) are still in early stages (Polling, 2021). ...
... Furthermore, Lycopodium spores and pollen grains have different morphological features that can influence predictive values. Some published pipelines incorporate a form of detection or localization of pollen-like objects that use techniques other than recall to assess model performance (e.g., Dell'Anna et al., 2009;Dunker et al., 2021;Maffezzoli et al., 2023;Mitsumoto et al., 2010;Pappas et al., 2003). For example, F-measures were reported in Dunker et al. (2021), FTIR spectra match values in Pappas et al. (2003) and correlation coefficients in Mitsumoto et al. (2010) that cannot directly be compared to our recall values. ...
... To the best of our knowledge, this is the first application of a fully automated pollen identification pipeline to develop stratigraphic data from fossil pollen samples (Bourel et al., 2020;Holt et al., 2011;Maffezzoli et al., 2023). Therefore, comparison is only possible with pipelines developed for the analysis of modern pollen assemblages or reference samples (e.g., Olsson et al., 2021;Oteros et al., 2020). ...
The automation of fossil pollen analysis promises many advantages in handling large numbers of samples with less resource allocation. However, automation is often obstructed by the high abundance of organic and min-erogenic non-pollen debris in fossil pollen samples. We used a Convolutional Neural Network-based approach to detect pollen-like objects in digital images of prepared microscopic slides for fossil pollen analysis and subsequently classified them into nine pollen classes and the marker spore Lycopodium. We trained the object detection and the classification model independently with a newly developed dataset of annotated images of fossil pollen grains. The object detection model achieved average recall rates of 89.8 % and 75.5 % for pollen classes and Lycopodium, respectively. The classification model correctly categorizes fossil pollen images with >95 % accuracy. We applied the assembled pipeline to Late Glacial pollen samples using class-dependent thresholds to discriminate true pollen from non-pollen objects and compared automated count data for nine pollen types with manual pollen counts. For the selected pollen types, our results demonstrate the feasibility to replicate major fossil pollen changes with automated counts, even when the automated pipeline was applied to pollen samples from a different site than used to train the models. High correlations (r = 0.97) between the first two axes of Principal Component Analyses (PCA) calculated based on automated and manual counts and high correlation (r = 0.93) indicated by a Procrustes rotation analysis of the PCA results demonstrate that the two procedures reconstructed similar pollen patterns. While our automated approach is not yet able to achieve the taxonomic resolution of manual counts by expert analysts and is limited to selected pollen types, it provides a "proof of principle" that automated analyses can be applied to complex fossil pollen samples and to develop downcore stratigraphies. Automated analyses may with time lead to reliable pollen records. For instance, our pipeline can be further improved by adding more pollen classes, increasing the dataset of annotated images of fossil pollen grains, expanding the training data to rare pollen types, refining taxonomic resolution (e.g., separation of Betula nana-type or Pinus-types), and incorporating more challenging pollen types (e.g., Juniperus), to expand its application beyond reconstructing temporal changes in a few selected pollen types.
... Alternatively, Al, Ti, and Rare Earth Elements concentrations have been interpreted as dust proxies (Sato et al., 2013). Recently, machine-learning techniques have been proposed to identify various classes of particles in ice cores based on optical images (Maffezzoli et al., 2023). ...
Detailed quantification of volcanic glass is crucial for improving the resolution of paleoenvironmental reconstructions and facilitating more accurate comparisons between distant sedimentary cryptotephra records. Here we present and evaluate two methods for the quantification of cryptotephra, shown on lake sediments from a site with distant Laacher See tephra fallout. Our methods initiate with delineating the extent and distribution of the cryptotephra layer within the sediments, accomplished through the integration of X-ray fluorescence (XRF) and computed tomography (medical- and μ-CT). The first quantification method involves the well-established process of shard extraction through stepwise density separation, followed by improved and statistically evaluated quantification introducing a new standardized marker. While the method itself is used widely for many years among cryptotephra researchers, we demonstrate how the new marker improves its precision for cryptotephra quantification, providing a robust, straightforward laboratory-based technique. Additionally, we introduce an innovative, software-based method that combines an SEM-based automated mineralogy analysis on thin sections with customized image analysis, which allows to study the area fraction of the glass phase, its depth-dependent variation, particle concentration with a focus on clustering behavior, depth-dependent particle count, total particle count, and particle size distribution within the glass phase. The significance of both methods lies in the efficiency and precision of cryptotephra quantification, enabling a deeper understanding of shard concentration and distribution. This study emphasizes the methodological innovations, offering improved tools for cryptotephra quantification, without focusing on detailed application-based analyses.
https://authors.elsevier.com/a/1k%7Eoo6fS6-nfyM
The pattern of atmospheric and climate changes recorded by coastal Antarctic ice core sites, and the processes they illustrate, highlight the importance of multiproxy studies on ice cores drilled from such peripheral areas, where regional to local-scale processes can be documented. Here, we present a 2000 year long record of aeolian mineral dust and diatoms windblown to the Roosevelt Island obtained from the RICE (Roosevelt Island Climate Evolution project) ice core. Mineral dust and diatoms are highly complementary at RICE since they are related to the large-scale South Pacific atmospheric circulation regime, carrying dust-rich air masses that travelled above the marine boundary layer, and local oceanic aerosol transport by low-level marine air masses, respectively. The 550–1470 CE period is characterized by enhanced mineral dust transport originating from the Southern Hemisphere continents, reduced sea-ice extent in the Eastern Ross and Amundsen Seas, and more frequent penetration of humid air masses responsible for the relative increase in snow accumulation. Around 1300 CE, in particular, in concomitance with marked El Niño-like conditions, the Ross Sea dipole reaches its maximum expression. After 1470 CE, relatively lower dust and snow deposition at RICE suggests an increase in pack ice. This period is characterized by episodes of unprecedented peaks of aeolian diatom deposition, indicating a rapid reorganization of atmospheric circulation linked to the eastward enlargement of the Ross Sea polynya, likely culminating with the opening of the proposed Roosevelt Island polynya, and to an increased influence of low-level marine air masses to the site during the Little Ice Age.
Ice cores of polar regions (ice sheets) are one of the most prominent natural archives that can reveal essential historical information from the past environment of our planet. The ice core microstructure is a key feature in determining the principal properties of ice such as pore close-off, albedo, and melt events. Micro-CT scans can provide valuable information about the microstructure of materials, although achieving a high-quality automated segmentation of porous materials, especially with phase/density changes is still a challenge. This work proposes a new method for improving the segmentation of porous microstructures where a weak segmentation (Gaussian Mixture Model) on high-resolution (30 μm) data is used as ground truth to train a deep learning model (U-net) for segmentation of low-resolution (60 μm) data. This approach has reached high segmentation accuracy in terms of quantitative metrics having the F1-score of 92.5% and an intersection over the union of 91%, with a considerable improvement compared to thresholding and unsupervised methods. Also, the segmentation results of U-net are closer to the real weight, density, and specific surface area of the specimen.
