The Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD) are invaluable tools to get physical insights into transient flows. They can be applied to extract spatial-temporal features that represent properties of the underlying dynamical system. As the computation of such decompositions can be quite expensive and requires the processing of a large amount of data, parallel and streaming (in-situ) algorithms for POD and DMD have recently become popular to save data in- and output during run-time, see e.g. [1]. However, before the simulation, a data sampling strategy and overall time interval must be defined, which can strongly influence the results. Recently, Rezaeiravesh et al. [2] introduced a framework to quantify the uncertainty in the sample mean estimator (SME) during simulation run-time. It models the auto-covariance function (ACF) at any spatial point in the domain, which can be used for an online estimate of the uncertainty due to finite time sampling. Based on this framework and the similarity between the Discrete Fourier Transform (DFT) and the DMD, we introduce a unique approach to monitor the convergence of DMD modes up to a predefined frequency. It can be utilized as a criterion to stop the simulation when a certain accuracy is reached, thus saving computing resources. Different sampling strategies and flow-quantity selections are compared for their influence on convergence and suitability for extracting DMD modes. Finally, selecting the dominant DMD modes for consecutive analysis steps can be tricky. Here, we introduce an adapted criterion from [3] that focuses on the use of modal structures for exploratory purposes. The results will be demonstrated using data from Large-Eddy-Simulations (LES) of the turbulent flow around two cylinders with varying relative positions at a Reynolds number of Re = 1.2*10^4. It contains several complex fluid phenomena, such as shear-layer interactions and flow separation and reattachment, that make it a demanding test case for investigating modal structures.