We simulate trajectories of energetic particles from impulsive solar flares for 2D+slab models of magnetic turbulence in spherical geometry to study dropout features, i.e., sharp, repeated changes in the particle density. Among random-phase realizations of two-dimensional (2D) turbulence, a spherical harmonic expansion can generate homogeneous turbulence over a sphere, but a 2D fast Fourier transform (FFT) locally mapped onto the lateral coordinates in the region of interest is much faster computationally, and we show that the results are qualitatively similar. We then use the 2D FFT field as input to a 2D MHD simulation, which dynamically generates realistic features of turbulence such as coherent structures. The magnetic field lines and particles spread nondiffusively (ballistically) to a patchy distribution reaching up to 25 degrees from the injection longitude and latitude at $r \sim 1$ AU. This dropout pattern in field line trajectories has sharper features in the case of the more realistic 2D MHD model, in better qualitative agreement with observations. The initial dropout pattern in particle trajectories is relatively insensitive to particle energy, though the energy affects the pattern's evolution with time. We make predictions for future observations of solar particles near the Sun (e.g., at 0.25 AU), for which we expect a sharp pulse of outgoing particles along the dropout pattern, followed by backscattering that first remains close to the dropout pattern and later exhibits cross-field transport to a distribution that is more diffusive, yet mostly contained within the dropout pattern found at greater distances.