Detection of changes in environmental phenomena using remotely sensed data is a major requirement in the Earth sciences, especially in natural disaster related scenarios where prompt detection, ideally in real-time, plays a crucial role in the saving of human lives and the preservation of natural resources. Although various approaches formulated to model multidimensional data can in principle be applied to the inherent spatial-temporal-spectral complexity of remotely sensed geospatial data, there are still challenging peculiarities that demand a precise characterization in the context of environmental change detection, particularly in scenarios of fast and abrupt changes. Similarly, although advances in data stream management systems (DSMS) could potentially be used in this case, such an approach mainly deals with tuple-based streams. Geospatial image streams, however, do not fit appropriately in the standard DSMS approach. In particular, the heterogeneous geographic coverage patterns followed by airborne and spacecraft sensors, along with the multidimensionality of the data in the streams, require more complex