Partitioning inter-well tracer test (PITT) is a direct method for quantifying remaining oil saturation (Sor), but conventional interpretation workflows can lead to significant overestimation if neglecting reservoir heterogeneity and mobile oil. This research introduces an advanced numerical workflow that overcomes these limitations by integrating the ensemble smoother with multiple data assimilation (ESMDA) with the UTCHEM reservoir simulator. UTCHEM is used to simulate production rate and tracer profile of each geological realization and ESMDA iteratively refines spatial distributions of both permeability (k) and Sor by jointly assimilating production and tracer data simultaneously. The workflow was successfully validated on synthetic reservoir models featuring high-permeability channels and spatially variable Sor. To demonstrate its superiority, the results were compared against both a traditional analytical solution and an assimilation case that excluded tracer data. Furthermore, a permeability-weighted Sor is proposed to better identify and prioritize oil recovery targets within the reservoir. The results demonstrate that coupling production and tracer data significantly cterization, reducing the estimation error by 86% compared to traditional analytical solutions. Our method yielded 3% error in Sor estimation compared to 22%–27% error by the traditional analytical methods. It accurately reconstructs the high-resolution spatial distribution of both k and Sor, provides an ensemble of solutions for uncertainty quantification, and is robust against data noise. This ensemble-based workflow offers a powerful technique for improving inter-well characterization, essential for both enhanced oil recovery and the assessment of non-aqueous phase liquid (NAPL) contamination.