Clustering algorithms partition a dataset into groups of similar points. The\nprimary contribution of this article is the Multiscale Spatially-Regularized\nDiffusion Learning (M-SRDL) clustering algorithm, which uses\nspatially-regularized diffusion distances to efficiently and accurately learn\nmultiple scales of latent structure in hyperspectral images. The M-SRDL\nclustering algorithm extracts clusterings at many scales from a hyperspectral\nimage and outputs these clusterings' variation of information-barycenter as an\nexemplar for all underlying cluster structure. We show that incorporating\nspatial regularization into a multiscale clustering framework results in\nsmoother and more coherent clusters when applied to hyperspectral data,\nyielding more accurate clustering labels.\n