A recent experiment has confirmed the existence of quantized turbulence in superfluid 3 He-B and suggested that turbulence is inhomogenous and spreads away from the region around the vibrating wire where it is created. To interpret the experiment we study numerically the diffusion of a packet of quantized vortex lines which is initially confined inside a small region of space. We find that reconnections fragment the packet into a gas of small vortex loops which fly away. We determine the time scale of the process and find that it is in order of magnitude agreement with the experiment. PACS 67.40. Vs, 47.37.+q The decay of superfluid turbulence at very low temperatures raises the fundamental question of the existence near absolute zero of an energy cascade from large to small scales. In the case of ordinary turbulence, the nonlinear terms of the Navier -Stokes equation distribute the energy over various scales without changing the total energy. The process leads to Richardson's cascade of bigger eddies breaking up into smaller eddies, until the wavenumber is large enough that kinetic energy is dissipated by viscous forces. In the case of superfluid turbulence, recent work indicates that generation of sound plays the role of 'sink' of kinetic energy [1]. Furthermore it appears that the generation of small scales occurs via creation of helical waves [2] of higher and higher wavenumbers on the quantized vortex filaments (Kelvin waves), and via creation of small vortex loops [3]. The nonlinear mechanism behind both processes is vortex reconnection, either indirectly (reconnections create cusps which relax into large amplitude Kelvin waves) or directly (reconnections create small vortex loops).The aim of this Letter is to show that the formation of small vortex loops is particularly important if the turbulence is not homogeneous, a case which is less investigated than homogeneous turbulence but is relevant to experiments performed at the lowest temperatures. In the case of 4 He many experiments [4] have been performed above 1.3K but much less is known about lower temperatures. In the most relevant study [5] turbulence was produced by oscillating a grid at temperatures as low as 20mK (0.01T c where T c is the critical temperature). Although there is no direct evidence, it is reasonable to expect that the vortex tangle was localized in the region of the grid. The detection was based on trapping ions on the quantized vortices, and measurements of the collected charge indicated that the total amount of turbulence in the cell decayed in time. In the case of 3 He-B, direct studies of vortices are typically done in a rotating cryostat [6], and an indirect observation of turbulence [7] [8] has been confirmed only recently [9]. In this experiment turbulence was created by vibrating a small wire shaped like a half-circle at temperatures around 0.11T c . Again, we expect that turbulence was localized in the region of the vibrating wire. Additional wires were used to detect the Andreev reflection of quasi-particles from the tu...