Abstract: A large number of openpit coal mines are distributed in the grasslands of north China, where soil and water erosion of the mining slope is serious and the resistance of ecological restoration is relatively high. In order to design the slope of the dump in the mine which is in line with the layout of surrounding natural grassland vegetation, we proposed a method of near-nature planting based on object learning. Taking the typical natural grassland around Shengli mining area in Inner Mongolia as an example, combined with the CLiDE evolution model, we evaluated soil antierosion effects of slope samples under different plant arrangements. The results showed that: (1) In terms of visual effects, the near-natural simulation results of plant layout based on object learning were similar to natural grassland, and both had stripe distribution characteristics at the mesoscale and random distribution characteristics at the small scale. (2) The statistical results showed that the simulation results of the near-natural layout were highly consistent with natural grassland in the number of cluster centers, distribution form, and the characteristic parameters of plant strips. (3) Under the same plant density, natural grassland slope had the strongest resistance of soil erosion, and the near-natural plant placement method can achieve simulation results that were very similar to the natural grassland in the slope resistance to water erosion. Moreover, compared with the traditional “uniform grid” plant layout under the same plant layout density, the nearnatural plant layout can reduce soil and water erosion by 46.3%. (4) In the restoration of slope vegetation in the grassland mining area, it can be considered to imitate plant distribution pattern of the surrounding natural grassland slope to improve the slope resistance to water erosion after restoration.

Key words: grassland, plant distribution, slope maintenance, Perlin noise, near-natural restoration.