Unmixing highly mixed grain size distribution data via maximum volume constrained end member analysis

TL;DR

This paper introduces MVC-EMA, a novel end member analysis method that effectively unmix highly mixed grain size distribution data by maximizing the volume of end members, improving the identification of true EMs.

Contribution

The paper proposes MVC-EMA, a new constrained EMA algorithm with a uniqueness theorem and quadratic programming, enhancing unmixing accuracy in highly mixed GSD data.

Findings

MVC-EMA accurately identifies true end members in highly mixed data.

The method outperforms existing EMA algorithms in complex sediment datasets.

Experimental results validate the effectiveness of MVC-EMA in sediment provenance analysis.

Abstract

End member analysis (EMA) unmixes grain size distribution (GSD) data into a mixture of end members (EMs), thus helping understand sediment provenance and depositional regimes and processes. In highly mixed data sets, however, many EMA algorithms find EMs which are still a mixture of true EMs. To overcome this, we propose maximum volume constrained EMA (MVC-EMA), which finds EMs as different as possible. We provide a uniqueness theorem and a quadratic programming algorithm for MVC-EMA. Experimental results show that MVC-EMA can effectively find true EMs in highly mixed data sets.