Use of principal component analysis for identification of Rockland and Trego Hot Springs tephras in the Hat Creek Graben, northeastern California, USA Solène Pouget a, ⁎, Marcus Bursik a , Joaquín A. Cortés a , Chris Hayward b a Department of Geology, University of Buffalo, SUNY, Buffalo, NY 14260, USA b School of Geosciences, University of Edinburgh, The King's Buildings, West Mains Road, Edinburgh EH9 3JW, UK abstract article info Article history: Received 15 May 2013 Available online xxxx Keywords: Tephrochronology Tephra dispersal Glass shard geochemistry Principal component analysis Burney Spring Mountain Electron microprobe Rockland Trego Hot Springs Hat Creek graben California Discontinuous tephra layers were discovered at Burney Spring Mountain, northern California. Stratigraphic relationships suggest that they are two distinct tephras. Binary plots and standard similarity coefficients of electron probe microanalysis data have been supplemented with principal component analysis to correlate the two tephra layers to known regional tephras. Using principal component analysis, we are furthermore able to bound our uncertainty in the correlation of the two tephra layers. After removal of outliers, within the 95% prediction interval, we can say that one tephra layer is likely the Rockland tephra, aged 565–610 ka, and the second layer is likely from Mt. Mazama, the Trego Hot Springs tephra, aged ~ 29 ka. In the case of the Rockland tephra, the new findings suggest that dispersal to the north was highly restricted. For Trego Hot Springs ash, the new findings extend the distribution to the southwest, with a rapid thinning in that direction. Coupled with considerations of regular tephra dispersal patterns, the results suggest that the primary dispersal direction for both tephras was to the south, and that occurrences in other directions are unlikely or otherwise anomalous. © 2013 University of Washington. Published by Elsevier Inc. All rights reserved. Introduction Tephra layers from volcanic plumes represent perhaps the most important geologic indicator of synchroneity in Earth history. Analyzing tephra samples has proven useful not only to chronostratigraphy, but also to the understanding of characteristics of past eruptions, such as their mass loading or their intensity (Shane, 2000). Tephra layers fine and thin with distance from the source (Walker and Croasdale, 1971). In the proximal region, fining and thinning are used as the two main diagnostic features to correlate the layer from one site to another (Rogova et al., 2007). Other common diagnostic lithostratigraphic features include grading, zoning, bedding and componentry. In more distal regions, where the tephra becomes thinner and where outcrops tend to contain layers from other eruptions and other volcanic sources, lithostratigraphic features become less diagnostic, while the geochemical signature increases in importance. Although geochemical analysis can yield a large amount of precise compositional data that is helpful in characterizing a layer, the data need to be treated carefully. First, geologic and chronologic context need to be considered in correlation. The geologic context includes not only characteristics and ages of overlying and underlying units and geomorphic setting, but also the volcanic context, which requires that tephra dispersal patterns implied by a given dataset make sense, as primary tephra layers thin and fine in regular patterns with distance from vent (Pyle, 1989; Fierstein and Nathenson, 1992; Sparks et al., 1992). Second, accepted statistical techniques need to be applied correctly for a more robust approach to analyzing the geochemical data to obtain reliable, quantitative correlation results, especially when the data are more challenging. One standard statistical technique that has been used with some success is Principal Component Analysis (PCA) (Pollard et al., 2006; Chiasera and Cortés, 2011), in which bivariate plots containing the vast majority of the statistical mass can be readily constructed and interpreted. The region of northeastern California between Medicine Lake volcano, Lassen volcanic center and Mt Shasta has not been known for accumulations of tephra, despite being surrounded by many potential sources (Sarna-Wojcicki et al., 1987; Rieck et al., 1992; Benson et al., 1997). During a field campaign to assess the volcanic stratigraphy of Burney Spring Mountain, an early Pleistocene stratovolcano 60 km NNE of Lassen Peak, California, discontinuous tephra layers were found during trenching (Fig. 1). In the present contribution, we describe the geochemistry of the tephras and using these data coupled with lithostratigraphic considerations, propose possible correlations and analyze potential dispersal patterns. Quaternary Research xxx (2013) xxx–xxx ⁎ Corresponding author at: Department of Geology, 411 Cooke Hall, University of Buffalo, Buffalo, NY, 14260, USA. E-mail address: solenepo@buffalo.edu (S. Pouget). YQRES-03497; No. of pages: 13; 4C: 0033-5894/$ – see front matter © 2013 University of Washington. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.yqres.2013.10.012 Contents lists available at ScienceDirect Quaternary Research journal homepage: www.elsevier.com/locate/yqres Please cite this article as: Pouget, S., et al., Use of principal component analysis for identification of Rockland and Trego Hot Springs tephras in the Hat Creek Graben, northeastern California, USA, Quaternary Research (2013), http://dx.doi.org/10.1016/j.yqres.2013.10.012