Journal of Conference Abstracts

Origin of Mount St. Helens Dacites by Partial Melting of Underplated Cascades Basalts

J. D. Blundy¹ (jon.blundy@bristol.ac.uk) & J. E. Gardner² (James_Gardner@brown.edu)

¹ CETSEI, Department of Geology, University of Bristol, Bristol.

² Department of Geological Sciences, Brown University, Providence, RI.

We have analysed trace elements (Li, LILE, REE, HFSE) in phenocrysts, matrix glasses, and glass inclusions in plagioclase from five MSH dacite pumices (Wn, We, T, 1980) by ion-microprobe. Inclusion and matrix glasses have near-identical compositions for all non-volatile major and trace elements indicating no significant episode of fractionation occurred between entrapment and eruption. Glasses, which represent unequivocal liquid compositions, show depletion in HREE relative to associated basalts. This observation (from whole-rock data) has been previously used to invoke melting of a garnet-bearing source, possibly eclogite from the subducted slab. That interpretation was based on published partition coefficients from magma systems other than MSH. Analyses of phenocryst rims and coexisting matrix glasses provide D's for amphibole, plagioclase, orthopyroxene and oxides at the conditions of the MSH magma chamber (~220 MPa, ~900°C) and therefore directly relevant to dacite petrogenesis. D's for clinopyroxene were calculated from the model of Wood & Blundy (CMP, submitted).

Significantly we find that amphibole has D_HREE>=4, while ilmenite has D_HFSE»1 (e.g. D_Nb~50). These observations (i) remove the requirement for garnet as a host for HREE in the source, and (ii) show that HFSE contents of dacites are unreliable petrogenetic indicators. Modelling of trace element behaviour using published phase relations/proportions of water-undersaturated basalt at 800 MPa and ~950°C, with amphibole+plagioclase in the residuum, reproduces the major and trace element chemistry of the dacites. The source is considered to be Cascades basalts trapped at ~25 km depth n the sub-MSH crust, a proposal in keeping with geophysical evidence. Crucially we find that Sr/Y ratios are not fractionated during partial melting and can be used as an indicator of source. Both basalt types occurring near MSH are implicated: pre-1800 dacites (Wn, We) were derived from Cave basalt, later dacites were derived from North Flank basalt. There is no evidence for slab melting at MSH.

Thickness Variations and Volume Estimates of Tephra Fall Deposits: the Importance of Particle Reynolds Number

C. Bonadonna¹ (fax: +39 50 500 675), G. G. J. Ernst² (Gerald.J.Ernst@bristol.ac.uk) & R. S. J. Sparks² (Steve.Sparks@bristol.ac.uk)

¹ Geology Dept, Pisa University, Pisa, Italy.

² Geology Dept, Bristol University, Bristol, UK.

Well-preserved tephra fall deposits display thickness variations which are more complex than simple exponential thinning. On plots of log thickness against square root of isopach area, many deposits show two or more straight-line segments and in some cases regions of curvature. We show that major changes in thinning rate occur as the particle size decreases with distance from the vent as a consequence of the change in settling behaviour from high to low Reynolds number as predicted by Rose (1993). Computer models of sedimentation from laterally spreading plumes predict a steep proximal region with exponential thinning for coarse ejecta (lapilli and coarse ash) with high Reynolds number (Re>500) and a distal region with power law thinning for fine ejecta settling at intermediate (500<Re<0.4) and low (Re<0.4) Reynolds number. Many prehistoric tephra deposits are not preserved beyond the more proximal high Reynolds number segment and deposit volume, using a simple exponential thinning law, is substantially underestimated when a large amount of ash is erupted. The data base (number of defined isopach contours) is usually too sparse to recognise the intermediate region so that the data will often be interpreted as two straight-line segments. The model predictions of the intersection of the high and low Reynolds number segments for different column heights show good agreement with observed breaks-in-slope. Data indicate that the decay constant (the thickness half-distance bt) for the proximal segment is correlated with column height as predicted by the models and is only weakly dependent on total grain size distribution. The distance to the distal break-in-slope is strongly correlated with column height and this is the basis for a new practical method for reconstructing column height from deposit dispersal.