Interferometric Synthetic Aperture Radar is commonly used in Earth Sciences to study surface displacements or construct high resolution topographic maps. Recent satellites such as Sentinel-1 constellation allow to derive dense deformation maps with millimetric precision thanks to high revisit frequency. However, InSAR is still limited by interferometric coherence. Interferogram phase noise resulting from a loss of coherence, due to changes in scattering properties between two SAR acquisitions, may lead to unwrapping errors. Unwrapping errors may lead to centimetric errors in time series reconstruction and in the estimation of ground velocity. We present an algorithm based on phase closure of triplets of interferograms to automatically correct unwrapping errors. We describe the algorithm and highlight its performances with two case studies, in Lebanon with Envisat satellite data and in Central Turkey with Sentinel-1 data. The first dataset is particularly affected by unwrapping errors because of long spatial and temporal baseline interferograms and decorrelation due in particular to the presence of vegetation. The second dataset contains unwrapping errors because of temporal changes in the scattering properties of the ground. For these two examples, the algorithm allows the correction of almost all unwrapping errors, without requiring visual inspection or manual deletions, hence preserving regions of interferograms where the phase is polluted by unwrapping errors. We illustrate the efficiency of the algorithm on large datasets (e.g. ``big data" problem), such as with Sentinel-1 constellation, where triplets redundancy enhances performances. Finally, we illustrate the influence of our method on time series reconstruction, removing inconsistencies in the derived velocity field.