We present a study of the long-term evolution of the cloud of aerosols produced in the atmosphere of Jupiter by the impact of an object on 19 July 2009 (Sánchez-Lavega, A. et al. [2010]. Astrophys. J. 715, L155-L159). The work is based on images obtained during 5 months from the impact to 31 December 2009 taken in visible continuum wavelengths and from 20 July 2009 to 28 May 2010 taken in near-infrared deep hydrogen-methane absorption bands at 2.1-2.3 μm. The impact cloud expanded zonally from ˜5000 km (July 19) to 225,000 km (29 October, about 180° in longitude), remaining meridionally localized within a latitude band from 53.5°S to 61.5°S planetographic latitude. During the first two months after its formation the site showed heterogeneous structure with 500-1000 km sized embedded spots. Later the reflectivity of the debris field became more homogeneous due to clump mergers. The cloud was mainly dispersed in longitude by the dominant zonal winds and their meridional shear, during the initial stages, localized motions may have been induced by thermal perturbation caused by the impact's energy deposition. The tracking of individual spots within the impact cloud shows that the westward jet at 56.5°S latitude increases its eastward velocity with altitude above the tropopause by 5-10 m s -1. The corresponding vertical wind shear is low, about 1 m s -1 per scale height in agreement with previous thermal wind estimations. We found evidence for discrete localized meridional motions with speeds of 1-2 m s -1. Two numerical models are used to simulate the observed cloud dispersion. One is a pure advection of the aerosols by the winds and their shears. The other uses the EPIC code, a nonlinear calculation of the evolution of the potential vorticity field generated by a heat pulse that simulates the impact. Both models reproduce the observed global structure of the cloud and the dominant zonal dispersion of the aerosols, but not the details of the cloud morphology. The reflectivity of the impact cloud decreased exponentially with a characteristic timescale of 15 days; we can explain this behavior with a radiative transfer model of the cloud optical depth coupled to an advection model of the cloud dispersion by the wind shears. The expected sedimentation time in the stratosphere (altitude levels 5-100 mbar) for the small aerosol particles forming the cloud is 45-200 days, thus aerosols were removed vertically over the long term following their zonal dispersion. No evidence of the cloud was detected 10 months after the impact.