Marine accumulations of terrigenous sediment are widely assumed to accurately record climatic- and tectonic-controlled mountain denudation and play an important role in understanding late Cenozoic mountain uplift and global cooling. Underpinning this is the assumption that the majority of sediment eroded from hinterland orogenic belts is transported to and ultimately stored in marine basins with little lag between erosion and deposition. Here we use a detailed and multi-technique sedimentary provenance dataset from the Yellow River to show that substantial amounts of sediment eroded from Northeast Tibet and carried by the river’s upper reach are stored in the Chinese Loess Plateau and the western Mu Us desert. This finding revises our understanding of the origin of the Chinese Loess Plateau and provides a potential solution for mismatches between late Cenozoic terrestrial sedimentation and marine geochemistry records, as well as between global CO₂ and erosion records.

Despite recent advances in quantitative methods of detrital provenance analysis, there is currently no widely accepted method of unmixing detrital geochronology age distributions. We developed a model that determines mixing proportions for source samples through inverse Monte Carlo modeling, wherein mixed samples are compared to randomly generated combinations of source distributions, and a range of best mixing proportions are retained. Results may then be used to constrain a forward optimization routine to find a single best‐fit mixture. Quantitative comparison is based on the Kolmogorov‐Smirnov (KS) testDstatistic and Kuiper testVstatistic for cumulative distribution functions, and the Cross‐correlation coefficient for finite mixture distributions (probability density plots or kernel density estimates). We demonstrate the capacity of this model through a series of tests on synthetic data, and published empirical data from North America mixed in known proportions; this proof‐of‐concept testing shows the model is capable of accurately unmixing highly complex distributions. We apply the model to two published empirical data sets mixed in unknown proportions from Colombia and central China. Neither example yields perfect model fits, which provides a cautionary note of potentially inadequate characterization of source and/or mixed samples, and highlights the importance of such characterization for accurate interpretation of sediment provenance. Sample size appears to be a major control on mixture model results; small (n < 100) samples may lead to misinterpretation. The model is available as a MATLAB‐based stand‐alone executable (.exe file) graphical user interface.

The Tibetan Plateau acted as a cradle of adaptation to cold for Pleistocene megafauna.