CORDIS - EU research results

Combining hyperspectral luminescence imaging and mineralogical identification to date mixed-component samples from volcanic eruptions

Periodic Reporting for period 1 - HYPERLIGHT (Combining hyperspectral luminescence imaging and mineralogical identification to date mixed-component samples from volcanic eruptions)

Reporting period: 2015-09-01 to 2017-08-31

Volcanic hazards directly affect more than 10% of the world’s population, and identifying eruption frequencies are fundamental to risk management. Yet there are significant gaps in records of prehistoric volcanic cycles due to geochronological limitations. ‘Un-dateable’ regions are manifold, and include the Mexico City basin, Japan, and New Zealand. Hence, there is a clear need for the development and application of innovative dating methodologies as part of integrated research into pre-historic volcanic events.

Luminescence dating may be able to provide ages for a number of eruptions that cannot be dated so far. This technique exploits the ability of quartz, feldspar, and glass to record the amount of radiation to which they have been exposed. If the dose rate of the environment is also calculated, the age since the last resetting event or crystallization is the absorbed dose divided by the dose rate. Luminescence dating spans the key age range (ca. 1,000-300,000 years), and can potentially be applied to any subaerial volcanic rock. Since the publication of a pioneering study on volcanogenic minerals in 1973, direct dating of volcanogenic material has been met with mixed success. Multiple studies have shown that these minerals often suffer from unstable luminescence signals and therefore underestimate eruption ages. Some promising results have been achieved by varying stimulation parameters and detection wavelengths, but contradictory results are common: a technique that produces an accurate age for one sample may not work for a chemically dissimilar crystal of the same mineral. Additionally, physical separation of mineral types may be difficult or impossible, as with microscopic inclusions of feldspar and quartz in glass shards.

Spatially-resolved luminescence (SRL) is the next step in research on dating volcanic ejecta. It measures luminescence by using an ultra-sensitive scientific camera (‘EMCCD’) to record sequential images during stimulation. In this way, many grains can be measured at the same time, then individual emission regions can be identified during data analysis. SRL has three key advantages over standard multigrain or single grain luminescence techniques:
1. Measurement of consolidated samples
2. Simultaneous measurement by heating
3. Discrimination of signals from ‘mixed’ samples

These advantages are directly applicable to volcanic dating. By measuring SRLfrom mixed and consolidated samples, one can bypass mineral separation which has hampered previous studies. Luminescence ages can also be compared with data from other analyses subsequently performed on the same samples, in order to reject signals from minerals known to give erroneous ages. Characterisation of their luminescence properties is a necessary first step in this direction, as a thorough understanding of the possible emission features and their stabilities must guide the formulation of any dating procedure.

Project HYPERLIGHT was designed to develope the following objectives:
1. A widely-applicable multi-method approach for dating volcanic ejecta
2. Flexible software for analysing SRL data

This action coincided with the planned coring of the Lake Chalco Basin (Mexico Citz), which is expected to yield a record of almost 1 million years of volcanic activity. Preliminary coring yielded numerous ash layers, but almost nothing is known about the long term evolution of the volcanoes in this basin. Chronological data on eruptions is of fundamental significance to volcanic hazard planning in the area, and it is expected that the methods developed during this project will be of significant future value to the program.
This project involved two main research strands:
1. Characterization of luminescence signals via SRL from single grains of bulk tephra samples
2. characterization of the luminescence signal of olivines.

SRL characterization of tephra was conducted for multiple widespread marker tephras well-characterized with respect to their geochemical makeup and component minerals. Experiments involved repeated irradiation of minerals, in order to build up a luminescence signal, and then measurements of the luminescence signal in multiple wavelengths with a scientific camera. Unfortunately, experiments quickly showed that the optical setup of the system was not sufficiently sensitive above camera background. All further characterizations were carried out upon the thermoluminescence (TL) signal. Key observations from these hyperspectral TL measurements include:
• UV (330 nm) and UV/blue emissions (380 nm) were too dim to successfully record.
• Glass shards yield a low intensity, widespread red TL (620 nm) emission.
• Plagioclase crystals may emit strongly in blue, yellow (565 nm), and red wavelengths, with intensity spanning several orders of magnitude.
• The intensity and wavelength of plagioclase emission is variable both inter- and intra-sample.

Significant energy was spent in the pursuit of developing functional image processing software. Multiple investigations have shown that the most fundamental hurdle to be overcome for SRL dating is that light smearing due to the imperfect focussing effects of simple optical systems. While the strength of SRL lies in the ability to stimulate many grains at the same time, and discriminate specific regions in the post-measurement image processing, Simple optical systems cannot perfectly focus the signal light, which leads to aberrations in the signal. There are several major consequences of this fact:
1. Light emitted from a given grain forms a ‘halo’ around that grain, and the extent of the detectable halo depends on the intensity of the signal.
2. The imaged shape of a grain changes depending on the position of the grain with respect to the lens. Grains towards the edge of an image tend to be elongated in the radial direction, and less well focussed.
3. Images of the same grains collected at multiple emission wavelengths do not share the same aberration response, therefore they cannot be compared 1-to-1 without further processing.

A MATLAB/ImageJ pipeline was developed that could read in luminescence signal image sequences as numerical arrays and fit the resulting data. Similar techniques were used to characterise volcanogenic olivines but in summary it is shown that developing an accurate dating method for olivines will be difficult.
It is demonstrated that SRL imaging is applicable to tephra minerals and glass. The differences inemitted luminescence are apparent between volcanogenic minerals from multiple eruptions, and between different components within a single tephra. While accurate signal measurements are still not possible due to light convolution from multiple grains, it is expected that further research will solve this problem. Once this has been accomplished, the findings of this project can be used as a springboard for further investigations into bulk tephra dating. Findings have been communicated within the luminescence and wider geological community. Olivine characterisation experiments do not show initial promise for use as a luminescence dosimeter, but it is possible that further screening and application of different techniques, such as pulsed stimulation, might provide more auspicious results. The results achieved within this project will form the base for a forthcoming application investigating the volcanic history in the Mexico City Basin.
Luminescence dating schematic.
Example of EMCCD-detected TL emissions
Spatially resolved luminescence emissions