We study the multiscale dynamics of inertial particle clustering using three-dimensional DNS data of isotropic turbulence with inertial particles considering different Stokes numbers. We apply a tessellation-based technique and thus assign a volume to each point particle. The temporal rate of change of the volumes yields the divergence of the particle velocity (Oujia et al., 2020; Maurel-Oujia et al., 2024). We perform a multiresolution analysis of the divergence on unstructured discrete particle positions (Matsuda et al., 2022). Thus the multiscale clustering dynamics is assessed and the spatial scales where clustering formation and destruction are most active, is determined for the different Stokes numbers.