Acquired signal, in any field of science and engineering, is almost always influenced by the unwanted, yet unavoidable noise from different sources of interference, which places signal filtering and estimation at the heart of signal processing methods. In case of contamination of spherical signals by realizations of anisotropic noise processes, which are spherical analogs of non-stationary processes in the Euclidean domain, it is desirable to construct filters that adapt according to the statistics of the noise on the sphere. In this talk, I will present such signal filtering techniques using joint domain representations, enabled by the spatially localized spherical harmonic transform and the scale-discretized wavelet transform on the sphere.