COMPARISON OF TOP-DOWN AND BOTTOM-UP APPROACHES TO QUANTITATIVE SEDIMENT PROVENANCE MODELLING IN THE PALEOGENE NORTHERN ALTIPLANO BASIN
We apply both top-down and bottom-up mixture modelling approaches to detrital zircon U-Pb ages from sediments in the Paleogene Muñani Formation and Puno Group. The top-down approach uses a Monte Carlo model to fit well-characterized source age distributions to mixed basin samples. In contrast, the bottom-up approach employs non-negative matrix factorization (NMF) to deconvolve source distributions based on multiple measured basin samples. Whereas the top-down approach attributes 40 – 50% to geographically ambiguous sources and points to Eastern Cordilleran contributions within uncertainty of zero, the bottom-up approach reduces this ambiguity and suggests there is up 30% contributed from the Eastern Cordillera. The additional clarity provided by empirical NMF results suggests that sources for the top-down approach are missing or inadequately characterized. This is the first study to use a bottom-up approach to sediment source characterization in the central Andes. The discrepancies between the top-down and bottom-up approaches highlight the difficulty in discriminating sediment sources in the central Andes, but also mark a step forward in provenance analysis within this area and to reconciling the basin history with thermo-kinematic models.