摘要
The paper is focused on computer simulation of natural vegetation propagation across two selected disturbed sites. Two sites located in the different environments, the abandoned sedimentation basin of a former pyrite ore mine and the ash deposits of a power station, were selected to illustrate the proposed spatio-temporal model. Aerial images assisted in identifying and monitoring the progress in the propagation of vegetation. Analysis of the aerial images was based on varying vegetation coverage explored by classification algorithms. A new approach is proposed entailing coupling of a local dynamic model and a spatial model for vegetation propagation. The local dynamic model describes vegetation growth using a logistic growth approach based on delayed variables. Vegetation propagation is described by rules related to seed and its dispersal phenomena on a local scale and on the scale of outlying spreading. The disturbed sites are divided into a grid of microsites. Each microsite is represented by a 5 m x 5 m square. A state variable in each microsite indicates the relative vegetation density on a scale from 0 （no vegetation） to 1 （long-term maximum of vegetation density）. Growth, local vegetation propagation and the effects of outlying vegetation propagation in each cell are described by an ordinary differential equation with delayed state variables. The grid of cells forms a set of ordinary differential equations. The abandoned sedimentation basin and the ash deposits are represented by grids of 185 x 345 and 212 x 266 cells, respectively. A few case-oriented studies are provided to show various predictions of vegetation propagation across two selected disturbed sites. The first case study simulates vegetation growing without spatial propagations and delayed variables in the spatio-temporal model. The second and the third case studies extend the previous study by including local and outlying vegetation propagation, respectively. The fourth case study explores delayed impacts in the logistic growth term and the delayed outcome by vegetation propagation across the disturbed space. The performed case-oriented studies confirm the applicability of the proposed spatio-temporal model to predict vegetation propagation in short-term successions and to estimate approximate vegetation changes in long-term development. As a result, it can be concluded that remotely sensed data are a valuable source of information for estimates of model parameters and provide an effective method for monitoring the progress of vegetation propagation across the selected sites, spaces disturbed by human activities.
