ESTIMATION IN COMPOSITIONAL DATA ANALYSIS

Compositional data arise naturally in several branches of science,including chemistry,geology,biology,medicine,ecology and manufacturing design.In chemistry,these constrained data seem to occur typicallywhen raw data are normalized or when output is obtained from a constrained estimation procedure,suchas might be used in a source apportionment problem.It is important not only for chemists to be awarethat the usual multivariate statistical techniques are not applicable to constrained data,but also to haveaccess to appropriate techniques as they become available.The currently available methodology is dueprincipally to Aitchison and is based on log-normal models.This paper suggests new parametric andnon-parametric approaches to significantly improve the existing methodology.In the parametric setting,some recent work of Rayens and Srinivasan is extended and a practical regression model is proposed.In the development of the non-parametric approach,minimum distance methods coupled withmultivariate bootstrap techniques are used to obtain point and region estimators.     

Glacially crushed quartz grains in loess as indicators of long-distance transport from major European ice centers during the Pleistocene

Detailed mineralogical analysis of the fine and very fine sand fractions (63–250) in the Dreihausen Loess Sequence shows the presence of numerous glacially crushed grains, including a significant amout of quartz. Previous theories regarding the origin of loesses in central Germany held that all sediments were derived locally, following short-distance transport from nearby sandstone outcrops. New data presented here show that many quartz and feldspar grains were glacially crushed, weathered, crushed again, transported, and emplaced by aeolian processes at Dreihausen. Crushing features observed are compatible with transport in both continental and alpine glaciers; some clasts show the effects of abrasion by both water and ice transport, while others are partly rounded presumably by fluvial and/or aeolian processes. Stronger palewind systems during the Pleistocene presumably contributed to the transport and emplacement of glacial grains from ice centers in both the Alps and the Baltic areas, and also possibly from the Rhine Basin.     

A ROLE FOR COVARIANCE STABILIZATION IN THE CONSTRUCTION OF THE CLASSICAL MIXTURE SURFACE

Some well-established ideas from classical covariance stabilization and shrinkage estimation are extendedto the context of physical mixtures through the use of a model developed by Burdick and Rayens.A cross-validation technique is proposed and associated computational issues are addressed.The methodology isevaluated in the context of real polychlorinated biphenyl data.     

Quantitative relationships between numbers of fluvial cycles, bulk lithological composition and net subsidence in a Scottish Namurian basin

This detailed quantitative basin analysis of fluvial deposits in a subsiding Namurian structural basin is aimed at discovering underlying statistical relationships between numbers of fluvial cycles, bulk lithological composition and net subsidence which could be used as a background to future sophisticated computer simulation experiments and would also facilitate comparison with other ancient basins. The succession studied lies between two widespread marine bands, one of Arnsbergian (E₂) and one of Kinderscoutian (R₁) age, and is dominated by upward-fining cycles: 94% of the semi-cycles containing sand, here termed grain-size cycles, are upward-fining and the ratio of fine members (mudstone+siltstone) to coarse members ranges from 0·23 to 5·0. Trend-surface analysis reveals basement structures, which influenced sedimentation, including a Caledonoid graben. Correlation coefficient values and results of principal component analysis demonstrate that the numbers of rooty horizons and grain-size cycles, together with the total thicknesses of sandstone and of mudstone+siltstone, all tend significantly towards a linear relationship with the total thickness of strata and hence net subsidence. The average thickness of grain-size cycles tends towards an inverse linear relationship with net subsidence. This probably reflects the presence of stacked, relatively thin, channel-fills within persistent channel belts. These belts tend to be localized in the areas of greatest subsidence within the basin and follow courses basically similar to channels already discovered in the underlying deltaic sediments. The highest concentrations of mudstone+siltstone tend to lie on the flanks of the basin, but high local sand concentrations found at points on the margin where channels entered the basin from the NW and NE and left to the SW, effectively disrupt any significant relationship between net subsidence and the proportion of sand. Coal has been selectively preserved on the flanks of the basin and the number of rooty horizons is greatest in an area of low subsidence which lay somewhat remote from the main channel belts.     

The tail of the Storegga Slide: insights from the geochemistry and sedimentology of the Norwegian Basin deposits

Deposits within the floor of the Norwegian Basin were sampled to characterize the deposition from the Storegga Slide, the largest known Holocene‐aged continental margin slope failure complex. A 29 to 67 cm thick veneer of variable‐coloured, finely layered Holocene sediment caps a homogeneous, extremely well‐sorted, poorly consolidated, very fine‐grained, grey‐coloured sediment section that is >20 m thick on the basin floor. This homogeneous unit is interpreted to represent the uppermost deposits generated by a gravity flow associated with the last major Storegga Slide event. Sediments analogous to the inferred source material of the slide deposits were collected from upslope on the Norwegian Margin. Sediments sampled within the basin are distinguishable from the purported source sediments, suggesting that size sorting has significantly altered this material along its flow path. Moreover, the very fine grain size (3·1 ± 0·3 μm) suggests that the >20 m thick homogeneous unit which was sampled settled from suspension after the turbulent flow was over. Although the turbulent phase of the gravity flow that moved material out into the basin may have been brief (days), significantly more time (years) is required for turbid sediments to settle and dewater and for the new sea floor to be colonized with a normal benthonic fauna. Pore water sulphate concentrations within the uppermost 20 m of the event deposit are higher than those normally found in sea water. Apparently the impact of microbial sulphate reduction over the last ca 8·1 cal ka bp since the re‐deposition of these sediments has not been adequate to regenerate a typical sulphate gradient of decreasing concentration with sub‐bottom depth. This observation suggests low rates of microbial reactions, which may be attributed to the refractory carbon composition in these re‐deposited sediments.     

Eocene fan delta-submarine fan deposition in the Wagwater Trough, east-central Jamaica

ABSTRACT The Wagwater Trough is a fault-bounded basin which cuts across east-central Jamaica. The basin formed during the late Palaeocene or early Eocene and the earliest sediments deposited in the trough were the Wagwater and Richmond formations of the Wagwater Group. These formations are composed of up to 7000 m of conglomerates, sandstones, and shales. Six facies have been recognized in the Wagwater Group: Facies I-unfossiliferous massive conglomerates; Facies II—channelized, non-marine conglomerates, sandstones, and shales; Facies III-interbedded, fossiliferous conglomerates and sandstones; Facies IV—fossiliferous muddy conglomerates; Facies V—channelized, marine conglomerates, sandstones, and shales; and Facies VI—thin-bedded sheet sandstones and shales. The Wagwater and Richmond formations are interpreted as fan delta-submarine fan deposits. Facies associations suggest that humid-region fan deltas prograded into the basin from the adjacent highlands and discharged very coarse sediments on to a steep submarine slope. At the coast waves reworked the braided-fluvial deposits of the subaerial fan delta into coarse sand and gravel beaches. Sediments deposited on the delta-front slope were frequently remobilized and moved downslope as slumps, debris flows, and turbidity currents. At the slope-basin break submarine fans were deposited. The submarine fans are characterized by coarse inner and mid-fan deposits which grade laterally into thin bedded turbidites of the outer fan and basin floor.     

The vertical turbulence structure of experimental turbidity currents encountering basal obstructions: implications for vertical suspended sediment distribution in non‐equilibrium currents

Large roughness features, caused by erosion of the sea floor, are commonly observed on the modern sea floor and beneath turbidite sandstone beds in outcrop. This paper aims to investigate the effect of such roughness elements on the turbulent velocity field and its consequences for the sediment carrying capacity of the flows. Experimental turbidity currents were run through a rectangular channel, with a single roughness element fixed to the bottom in some runs. The effect of this roughness element on the turbulent velocity field was determined by measuring vertical profiles of the vertical velocity component in the region downstream of the basal obstruction with the Ultrasonic Doppler Velocity Profiling technique. The experiments were set up to answer two research questions. (i) How does a single roughness element alter the distribution of vertical turbulence intensity? (ii) How does the altered profile evolve in the downstream direction? The results for runs over a plane substrate are similar to data presented previously and show a lower turbulence maximum near the channel floor, a turbulence minimum associated with the velocity maximum, and a turbulence maximum associated with the upper flow interface. In the runs in which the flows were perturbed by the single roughness element, the intensity of the lower turbulence maximum was increased between 41% to 81%. This excess turbulence dissipated upwards in the flow while it travelled further downstream, but was still observable at the most distal measurement location (at a distance ca 39 times the roughness height downstream of the element). All results point towards a similarity between the near bed turbulence structure of turbidity currents and free surface shear flows that has been proposed by previous authors, and this proposition is supported further by the apparent success of a shear velocity estimation method that is based on this similarity. Theory of turbulent dispersal of suspended sediment is used to discuss how the observed turbulent effects of a single large roughness element may impact on the suspended sediment distribution in real world turbidity currents. It is concluded that this impact may consist of a non‐equilibrium net‐upwards transport of suspended sediment, counteracting density stratification. Thus, erosive substrate topography created by frontal parts of natural turbidity flows may super‐elevate sediment concentrations in upper regions above equilibrium values in following flow stages, delay depletion of the flow via sedimentation and increase their run‐out distance.