Correlating tephras and cryptotephras using glass compositional analyses and numerical and statistical methods: Review and evaluation

Glass composition-based correlations of volcanic ash (tephra) traditionally rely on extensive manual plotting. Many previous statistical methods for testing correlations are limited by using geochemical means, masking diagnostic variability. We suggest that machine learning classifiers can expedite correlation, quickly narrowing the list of likely candidates using well-trained models. Eruptives from Alaska's Aleutian Arc-Alaska Peninsula and Wrangell volcanic field were used as a test environment for 11 supervised classification algorithms, trained on nearly 2000 electron probe microanalysis measurements of glass major oxides, representing 10 volcanic sources. Artificial neural networks and random forests were consistently among the top-performing learners (accuracy and kappa > 0.96). Their combination as an average ensemble effectively improves their performance. Using this combined model on tephras from Eklutna Lake, south-central Alaska, showed that predictions match traditional methods and can speed correlation. Although classifiers are useful tools, they should aid expert analysis, not replace it. The Eklutna Lake tephras are mostly from Redoubt Volcano. Besides tephras from known Holocene-active sources, Holocene tephra geochemically consistent with Pleistocene Emmons Lake Volcanic Center (Dawson tephra), but from a yet unknown source, is evident. These tephras are mostly anchored by a highly resolved varved chronology and represent new important regional stratigraphic markers.

Section: Resultsmentioning

confidence: 99%

“…We agree with Lowe et al . () that robust correlations must consider some combination of physical, stratigraphic, compositional and chronologic parameters. In addition, we strongly suggest the use of plotting to assess input data before models are fit to ensure that only the best quality data are used.…”

Section: Resultsmentioning

confidence: 99%

Machine learning classifiers for attributing tephra to source volcanoes: an evaluation of methods for Alaska tephras

Bolton

Jensen

Wallace

et al. 2019

“…A hierarchical cluster analysis of the elemental data for the volcanic glass shards from the Suigetsu tephra layers was carried out to statistically evaluate the relationships amongst the Suigetsu tephra layers. Cluster analysis has been tested and used for evaluation of the relationships between tephras (e.g., Baron et al ., ; Kikkawa et al ., ; Lowe et al ., ) in tephrochronological studies.…”

Section: Samples and Analytical Methodsmentioning

confidence: 99%

Major and trace element abundances in volcanic glass shards in visible tephras in SG93 and SG06 drillcore samples from Lake Suigetsu, central Japan, obtained using femtosecond LA–ICP–MS

Maruyama

Takemura

Hirata

et al. 2019

The major and trace element concentrations of volcanic glass shards from visible tephra layers in the SG93 and SG06 cores from Lake Suigetsu, central Japan, were determined by femtosecond laser ablation–inductively coupled plasma–mass spectrometry. The glass‐shard analyses, together with the petrographic properties of the tephra samples, allow the Suigetsu tephra layers to be broadly classified into tephras derived from calderas on Kyushu Island, and from Daisen and Sambe volcanoes in the Chugoku district of southwest Japan. The layers correlated with tephras from Kuju caldera and Daisen volcano, and with the younger Sambe tephras, have adakitic elemental features. A Suigetsu tephra sample correlated with the Sambe−Kisuki tephra based on petrographic properties has an elemental pattern similar to that of the Toya tephra from Hokkaido Island, northeast Japan. This match implies that tephras from northeast Japan, as well as Kyushu–Chugoku tephras, are possible correlatives of the Suigetsu tephra layers. Both petrographic properties and major–trace element data of volcanic glass shards are essential for robust tephra correlations, and hierarchical cluster analysis proved additionally useful in statistically evaluating relationships among the tephras.

“…Both laser‐ablation inductively coupled‐plasma mass spectrometry (LA‐ICP‐MS) and electron probe microanalysis (EPMA) by scanning electron microscopy with energy‐dispersive X‐ray spectroscopy (SEM‐EDS) provide the advantage of analysing individual glass shards without the unfavourable incorporation of coexisting mineral grains. Moreover, LA‐ICP‐MS provides an efficient, accurate and suitably precise method for determining the abundances of a wide variety of trace elements, including rare‐earth elements (REEs) at low concentrations in individual glass shards (Pearce, ; Lowe et al ., ). In addition, femtosecond LA‐ICP‐MS, used in the present study, has the additional advantage of being able to obtain major element as well as trace element data together for glass (Maruyama et al ., ).…”

Section: Introductionmentioning

confidence: 98%

Identification of Lower Pleistocene widespread tephras associated with large caldera‐forming eruptions in the Tohoku area, north‐east Japan

Suzuki

Maruyama²,

Danhara³

et al. 2019

Tephra fingerprinting techniques contributing to volcanology and palaeoenvironmental studies have been developed using a combination of laser-ablation inductively coupled-plasma mass spectrometry (LA-ICP-MS) and scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDS). In particular, femtosecond LA-ICP-MS can determine major-and trace element abundances in individual glass shards. On the basis of the major oxide and trace element composition of the glass shards, using those methods, we re-examined the identification of four lower Pleistocene tephras originating from north-east Japan. All trace element abundances exhibited the typical pattern of tephras from the Hokkaido-Tohoku area, and major element concentrations were distinct. As a result, we re-examined the correlation of the widespread Tmg-R4 tephra (2.0 Ma), and newly defined the widespread Kd44-Naka tephra (1.968-1.781 Ma), both originating from the Sengan geothermal region. Furthermore, we re-examined identifications of Sr-Asn-Kd8 (1.219 Ma) and Sr-Kc-U8 (0.922-0.910 Ma) in central Japan, both derived from the Aizu volcanic region. The extensive distributions of the former two tephras suggest the occurrence of two large caldera-forming eruptions (Volcanic Explosivity Index 7) during a short period. Also, the distributions and volumes of the latter two tephras are broader and larger than those previously assumed. The results provide insight into large volcanic eruption history and terrestrial and marine palaeoenvironmental history.