Source Apportionment of Weathered Oil Spills

In oil pollution research, identifying spill sources is crucial but challenging due to mixed/weathered samples. This study simulated 300 days of controlled abiotic weathering on oil mixtures with two crude oils and analyzed the resulting isotopic signatures using GC/C/IRMS. Carbon and hydrogen isotope profile of mixed oil is mainly influenced by the weight proportion of different end-numbers. Six new oil end numbers are founded by different change modes of these two isotope values (carbon (δ13C) and hydrogen (δD) isotope compositions) of n-alkanes in mixed oil samples. The contributions of different end numbers can be quantified by δD values of Pr and Ph in the oil samples. Initial mixing ratios of the original crude oils were determined as 42.4%–26.9% and 66.3%–5.0%. By integrating non-negative matrix factorization (NMF) with gradient descent and coordinate descent algorithms, refined mixing ratios of 35.8%–28.6% and 64.0%–8.3% were derived, which strongly correlated with prior results and validated the algorithm accuracy. Further analysis revealed source resolution limits: for one original crude oil, carbon isotopes indicated a maximum of four-end-member mixing (38.0%–19.0%), consistent with hydrogen isotopes (40.4%–17.2%), suggesting a four-source cap. In contrast, the other crude oil tolerated up to five end-members under both carbon (38.7%–8.0%) and hydrogen (34.3%–6.6%) isotopic analyses, demonstrating a five-source resolution capacity. This study integrates isotopic analysis with computational algorithms, enhancing the accuracy of oil spill source attribution and clarifying the theoretical limits of source resolution in complex mixtures.