Periodic Reporting for period 1 - CREDit (Chronological REference Datasets and Sites (CREDit) towards improved accuracy and precision in luminescence-based chronologies)
Periodo di rendicontazione: 2020-01-01 al 2021-12-31
Generally, our project addressed the challenge of reproducibility of modern science. More specifically, the project's overall aim was to develop and implement new ways to improve the accuracy and precision of luminescence-derived chronologies in Quaternary sciences through reference datasets and samples.
Luminescence dating reaches back at least 250,000 years. In geological terms, nothing more than a blink of an eye. However, this period is of uppermost interest for understanding past terrestrial landscape dynamics, from which future landscape evolution crucial to modern societies can be derived.
Reliable and reproducible, luminescence-based chronologies are cornerstones to successfully integrating findings from various disciplines in Earth sciences (e.g. geomorphology, soil science, palaeoecology) and archaeology. Furthermore, it allows to match past climate patterns found in terrestrial archives (e.g. loess profiles or varves) and compare them with findings in ice and marine archives. Because of the success and popularity of luminescence dating to establish chronologies, every improvement in accuracy and precision positively impacts numerous studies building on such results. In the mid-and long-term, this situation generates a positive outcome for the society in terms of a better understanding of past landscape dynamics, leading to improved and more precise model predictions for future changes. Furthermore, improved results will automatically lower the costs of such research in the long term by minimising repetitions through better data quality.
On a discipline level, towards the broad overall goal of improved accuracy and precision of luminescence-based chronologies, we aimed to create reference data and samples (and tools creating such data), enabling comparisons of methods (e.g. statistical models) and results. We proposed to work with an annually layered sediment core, acting as a reference site and compile those data combined with artificially generated datasets.
To that end, CREDit proposed: (1) Sampling and luminescence dating of the core DE3 from the Eifel-Laminated-Sediment-Archive (ELSA), for which a known chronology exists (up to 20 samples), (2) design, test, and distribution of luminescence-based reference datasets, and (3) application of the obtained data, including comparing analysis tools to improve age-depth models. We proposed the statistical programming environment R for designing and testing the reference data.
Because of the pandemic, we had reshuffled our work packages and focused more on artificial reference data. With the publication process still ongoing, we were able to sample and analyse the varve core and developed various tools to dynamically generate reference data; a slight modification of the initial plan, where we had proposed static datasets (which are still possible). The developed tools to generate reference data dynamically are novel and will contribute to better luminescence-derived ages. However, to data is not yet clear whether the layer-sediment record can act as the proposed reference dataset because of unexpected (too old) age results.
Luminescence dating reaches back at least 250,000 years. In geological terms, nothing more than a blink of an eye. However, this period is of uppermost interest for understanding past terrestrial landscape dynamics, from which future landscape evolution crucial to modern societies can be derived.
Reliable and reproducible, luminescence-based chronologies are cornerstones to successfully integrating findings from various disciplines in Earth sciences (e.g. geomorphology, soil science, palaeoecology) and archaeology. Furthermore, it allows to match past climate patterns found in terrestrial archives (e.g. loess profiles or varves) and compare them with findings in ice and marine archives. Because of the success and popularity of luminescence dating to establish chronologies, every improvement in accuracy and precision positively impacts numerous studies building on such results. In the mid-and long-term, this situation generates a positive outcome for the society in terms of a better understanding of past landscape dynamics, leading to improved and more precise model predictions for future changes. Furthermore, improved results will automatically lower the costs of such research in the long term by minimising repetitions through better data quality.
On a discipline level, towards the broad overall goal of improved accuracy and precision of luminescence-based chronologies, we aimed to create reference data and samples (and tools creating such data), enabling comparisons of methods (e.g. statistical models) and results. We proposed to work with an annually layered sediment core, acting as a reference site and compile those data combined with artificially generated datasets.
To that end, CREDit proposed: (1) Sampling and luminescence dating of the core DE3 from the Eifel-Laminated-Sediment-Archive (ELSA), for which a known chronology exists (up to 20 samples), (2) design, test, and distribution of luminescence-based reference datasets, and (3) application of the obtained data, including comparing analysis tools to improve age-depth models. We proposed the statistical programming environment R for designing and testing the reference data.
Because of the pandemic, we had reshuffled our work packages and focused more on artificial reference data. With the publication process still ongoing, we were able to sample and analyse the varve core and developed various tools to dynamically generate reference data; a slight modification of the initial plan, where we had proposed static datasets (which are still possible). The developed tools to generate reference data dynamically are novel and will contribute to better luminescence-derived ages. However, to data is not yet clear whether the layer-sediment record can act as the proposed reference dataset because of unexpected (too old) age results.
As proposed during the reporting period, we sampled and analysed the core DE4 (technical amendment: core DE4, instead of core DE3) from the Dehner dry maar. Additionally, we developed new tools to generate artificial reference data dynamically, following a slight modification of our research approach, using the statistically programming environment R. Additionally, during the project, we realised that a data format to store luminescence data (including reference data) is missing, but needed to align with the FAIR guidelines for open science. We, therefore, proposed a new data format for long-term data storage and exchange of luminescence data.
The project has produced 5 publications in international-peer review journals, 3 scientific datasets, 3 R packages, and 3 conference contributions, and 1 invited talk. Three additional contributions for international peer-reviewed journals and two R packages are pending submission. Our submission timeline for the pending publications: July 2022 (new data format luminescence data, including reference data), autumn 2022 (results of the varve core dating), and end 2022 (final compilation of the reference data) for the last manuscript.
Unfortunately, the project had two setbacks: (1) the global pandemic delayed the start of the sampling of the varve core (countermeasure: reshuffling work package order), and (2) it appears that the varve core is unsuitable for providing reference samples, because the obtained luminescence ages are an order of magnitude older than expected and hence a meaningful comparison to the known ages seems questionable. Nevertheless, the samples may still provide valuable insight into the regional sedimentation history with implications for the European loess history.
Positively, the reshuffled work packages enabled us to focus more on the artificial reference datasets and delve deeper into the subject. For instance, we developed a new R package, ‘RLumCarlo,’ enabling the simulation of stochastic uncertainties stemming from the luminescence production process. Additionally, the package ‘sandbox’ opens the possibility of generating datasets using a 2D sediment model that considers sediment texture properties and links them to luminescence characteristics.
The project has produced 5 publications in international-peer review journals, 3 scientific datasets, 3 R packages, and 3 conference contributions, and 1 invited talk. Three additional contributions for international peer-reviewed journals and two R packages are pending submission. Our submission timeline for the pending publications: July 2022 (new data format luminescence data, including reference data), autumn 2022 (results of the varve core dating), and end 2022 (final compilation of the reference data) for the last manuscript.
Unfortunately, the project had two setbacks: (1) the global pandemic delayed the start of the sampling of the varve core (countermeasure: reshuffling work package order), and (2) it appears that the varve core is unsuitable for providing reference samples, because the obtained luminescence ages are an order of magnitude older than expected and hence a meaningful comparison to the known ages seems questionable. Nevertheless, the samples may still provide valuable insight into the regional sedimentation history with implications for the European loess history.
Positively, the reshuffled work packages enabled us to focus more on the artificial reference datasets and delve deeper into the subject. For instance, we developed a new R package, ‘RLumCarlo,’ enabling the simulation of stochastic uncertainties stemming from the luminescence production process. Additionally, the package ‘sandbox’ opens the possibility of generating datasets using a 2D sediment model that considers sediment texture properties and links them to luminescence characteristics.
The most notable achievement of the project so far is raising awareness in the luminescence-dating community about the challenges that require reference datasets in the first place to improve accuracy and precision and not, at last, the reproducibility of luminescence-based chronologies. One example is our most recent contribution to a publication, where two reference quartz samples are systematically compared to understand quartz luminescence production better, and it will lead to a repository of such data in the future. While this research is more a spin-off of our project, it yet leads to more accurate and precise luminescence-based chronologies due to a better understanding of the underlying processes. While last publications are still pending, a few outputs so far progressed beyond state of the art. For instance, the R package ‘RLumCarlo’ provides a new way to investigate stochastic uncertainties in luminescence data. These uncertainties, particularly for single grain measurements, were not yet considered. At the same time, ‘sandbox’ enables, for the first time, to combine sediment texture with luminescence characteristics to generate reference datasets dynamically and for a broader range of scenarios. In the latter publication, we could show through simulations that part of the age scatter in luminescence ages is attributed to the sampling (reduced precision). Our simulations will help develop better sampling strategies and at least quantify the expected effect on the age results.
For the upcoming publications, we expect a major impact with the newly proposed data format, because it will foster data exchange and transparency.
For the upcoming publications, we expect a major impact with the newly proposed data format, because it will foster data exchange and transparency.