Dynamic Mode Decomposition (DMD) is a powerful data-driven method used to\nextract spatio-temporal coherent structures that dictate a given dynamical\nsystem. The method consists of stacking collected temporal snapshots into a\nmatrix and mapping the nonlinear dynamics using a linear operator. The standard\nprocedure considers that snapshots possess the same dimensionality for all the\nobservable data. However, this often does not occur in numerical simulations\nwith adaptive mesh refinement/coarsening schemes (AMR/C). This paper proposes a\nstrategy to enable DMD to extract features from observations with different\nmesh topologies and dimensions, such as those found in AMR/C simulations. For\nthis purpose, the adaptive snapshots are projected onto the same reference\nfunction space, enabling the use of snapshot-based methods such as DMD. The\npresent strategy is applied to challenging AMR/C simulations: a continuous\ndiffusion-reaction epidemiological model for COVID-19, a density-driven gravity\ncurrent simulation, and a bubble rising problem. We also evaluate the DMD\nefficiency to reconstruct the dynamics and some relevant quantities of\ninterest. In particular, for the SEIRD model and the bubble rising problem, we\nevaluate DMD's ability to extrapolate in time (short-time future estimates).\n