Unsupervised change detection labels spectrally-similar groups and clusters at Date 1, followed by spectrally similar groups at Date 2, and then detects changes. Because an unsupervised algorithm is used, the process can be automated, but labelling the changed areas is not always straightforward, especially in the case of processes (i.e. a series of changes that are part of a process, such as conversion of forest via burnt areas to crops and, finally, pasture).