Dense seismic array monitoring combined with advanced processing can help retrieve and locate a variety of seismic sources with unprecedented resolution and spatial coverage. We present a methodology that goes beyond classical localization algorithms through gathering various types of sources (impulsive or continuous) using a single scheme based on a gradient‐descent optimization and evaluating different levels of phase coherence. We apply our methodology on an Alpine glacier and demonstrate that we can retrieve the dynamics of active crevasses with a metric resolution using sources associated with high phase coherence; the presence of diffracting materials (e.g., rocks) trapped in transverse crevasses using sources with moderate phase coherence; and the two‐dimensional time evolution of the subglacial hydrology system using sources with low phase coherence. Our study highlights the strength of using an appropriate and systematic seismological approach to image a wide range of subsurface structures and phenomena in settings with complex wavefields. Plain Language Summary: Over the past two decades, the growing use of dense seismic arrays has often overcome limitations of traditional observations methods and yielded new insights on the physics of subsurface process and properties. Yet scientific and computational challenges remain to be addressed for using the appropriate array‐processing approaches and automating the techniques on large volume of data and for complex wavefields. In this paper we address such challenges in the particular case of monitoring glaciers, which host numerous and diverse sets of seismic sources that produce signals ranging from impulsive to tremor‐like. We combine a physics‐based and a statistical approach to explore with a dense seismic array the spatial coherence of the seismic wavefield generated by such a diversity of sources. We show that even a small coherence in the phase signal remains rich in statistical information on concomitant and/or low amplitudes micro‐seismic sources. This allows us to localize seismic sources with a super‐resolution (meter to decameter) and identify emerging patterns associated with a wide range of glacier features and their dynamics, ranging from active crevasses, debris in transverse passive crevasses and subglacial water flow. Such methodological and conceptual advance may enable a more efficient and complete imaging of geophysical objects. Key Points: We present an innovative array‐processing approach to image glaciers structures at high resolution by locating seismic sourcesWe investigate a large range of spatial phase coherences, from very low up to very high, over narrow frequency bands and short time windowsWe image the spatial and temporal dynamics of sources originating from active and passive crevasses as well as from subglacial hydrology [ABSTRACT FROM AUTHOR]