Garnet chemistry provides a well-established tool in the discrimination and interpretation of sediment provenance. Current discrimination approaches, however, (i) suffer from using less variables than available, (ii) subjective determination of discrimination fields with strict boundaries suggesting clear separations where in fact probabilities are converging, and (iii) significant overlap of compositional fields of garnet from different host-rock groups. The new multivariate discrimination scheme is based on a large database, a hierarchical discrimination approach involving three steps, linear discriminant analysis at each step, and the five major host-rock groups to be discriminated: eclogite-(A), amphibolite-(B) and granulite-(C) facies metamorphic rocks as well as ultramafic (D) and igneous rocks (E). The successful application of statistical discrimination approaches requires consideration of the a priori knowledge of the respective geologic setting. This is accounted for by the use of prior probabilities. Three sets of prior probabilities (priors) are introduced and their advantages and disadvantages are discussed. The user is free to choose among these priors, which can be further modified according to the specific geologic problem and the level of a priori knowledge. The discrimination results are provided as integrated probabilities of belonging to the five major host-rock groups. For performing calculations and results a supplementary Excel® spreadsheet is provided. The discrimination scheme has been tested for a large variety of examples of crystalline rocks covering all of the five major groups and several subgroups from various geologic settings. In most cases, garnets are assigned correctly to the respective group. Exceptions typically reflect the peculiarities of the regional geologic situation. Evaluation of detrital garnets from modern and ancient sedimentary settings of the Western Gneiss Region (Norway), Eastern Alps (Austria) and Albertine Rift (Uganda) demonstrates the power to reflect the respective geologic situations and corroborates previous results. As most garnet is derived from metamorphic rocks and many provenance studies aim at reconstructing the tectonic and geodynamic evolution in the source area, the approach and the examples emphasize discrimination of metamorphic facies (i.e., temperature-pressure conditions) rather than protolith composition.
23Detrital heavy minerals commonly document the geological setting in the source area, hence they are 24 widely used in sedimentary provenance analysis. In heavy mineral studies most commonly the 63−125 25 and 63−250 µm grain-size fractions are used. Heavy mineral data and garnet geochemistry of stream 26 sediments and bedrocks from the catchment area draining the Almklovdalen peridotite massif in SW 27Norway reveal that a wider grain-size spectrum needs to be considered to avoid misleading 28 interpretations. The Almklovdalen peridotite massif consists mainly of dunite and harzburgite, as 29 testified by the heavy mineral suite. At the outlet of the main river the heavy mineral spectrum is very 30 monotonous due to dilution by strong influx of olivine. Heavy minerals like apatite and epidote 31 characterising the host gneisses have almost disappeared. MgO-rich almandine garnets are more 32 frequent in the coarser grain-size fractions, whereas MnO-rich almandine garnets are more frequent in 33 the finer grain-size fractions. Garnets with pyrope content exceeding 50 % are only found in the 34 500−1000 µm grain-size fraction. Therefore, the sample location and the selected grain-size fraction 35
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.