Atmospheric radioactive releases have long-term health and environmental impacts following nuclear accidents. Determining the temporal rate and total amount of the release during such events is challenging using forward physical models based on reactor conditions. The environmental measurements, e.g., air concentrations, dose rates, and surface deposition, record the transport process of the released radionuclides, acting as the indicator to trace the emissions. Following the Fukushima accident, inverse modeling technique has been used for the source term estimation by minimizing the simulation-to-observation discrepancies. Due to the limitations of single-source measurements, supplementary data from diverse sources is needed to compensate for the incomplete release information. However, existing methods rely on expert decision-making for subjective judgment and lack effective objective multi-source inversion approaches. In this study, we investigate the objective inversion using multi-source measurements for the 137Cs source term estimation following the Fukushima accident. Such inversions were implemented based on the framework with a joint correction function and a total variation regularization to account for the systematic model biases and the representation of piecewise-constant release characteristics. Three surface deposition datasets were collected for this task, as well as an atmospheric concentration dataset, which provided eight cases with different input data. The inversion results for these cases were compared and used for the atmospheric dispersion simulation for the evaluation against the environmental observations.