Grain Size Effect Of Vegetation Landscape Pattern In Saihanba Forest Farm

Landscape pattern analysis is one of the core contents of landscape ecology research and is of great significance for understanding ecological processes.Scale effect research is an important part of landscape pattern analysis.Analyzing the scale effects of vegetation landscape patterns and identifying their characteristic scales are helpful for accurately revealing the characteristics of vegetation landscape patterns,vegetation landscape patterns,and developing scientific and effective vegetation resources.Based on the theory of spatial scale of landscape ecology,this study analyzed the spatial-temporal scale effect of the vegetation landscape pattern in the Saihanba forest farm,and finally determined its characteristic scale.Based on the random forest classification method,three phases(1989,2000 and 2016)of remote sensing images in Saihanba forest farm were interpreted.Landscape type raster data was scaled up by simple aggregation method and landscape metrics were calculation based on patch type and landscape levels.The scale map was used to analyze the landscape scale effects of four types of vegetation,including grassland,natural secondary forest,shrub and plantation,at different spatial and temporal scales.The optimal spatial grain size of landscape pattern analysis was determined by the information entropy model.The main results were as follows:(1)The dynamic changes of land cover types in the study area were analyzed based on the results of the three phases of remote sensing images.The results showed that the four vegetation landscapes(grassland,shrub forest,plantation,and natural secondary forests)changed significantly during 1989-2016.The area decreased by 141.48km2 with the largest lowering extent(88.07%);the area of natural secondary forest decreased by 56.27km2 with a lowering extent of 18.83%;the area of plantation increased by 190.83km2 with the largest increasing extent(102.42%);the grassland area increased by 13.37km2 with an increasing extent of 5.39%.(2)Different landscape metrics responded differently to the spatiotemporal scale.Among them,patch type metrics(edge density,patch density,aggregation index and landscape type metrics(shape index,contagion index)had a strong scale dependence.Contrast to landscape type metrics(0.29-70.02%),the patch type metrics(7.79-96.70%)generally had a stronger scale effect.(3)The four vegetation landscapes,including grassland,natural secondary forest,shrub and plantation)had the same changing characteristics on the spatial scale,The scale effect continued to weaken with the increase of spatial grain size.At the temporal scale,the edge density showed a decreasing trend in shrub forest and natural secondary forest,and showed an increasing trend in grassland and plantation.Patch density showed a decreasing trend in shrub forest and plantation,and increased in grassland and natural secondary forest.Aggregation index and the maximum patch index both decreased in grassland and shrub forest,and increased in plantation and natural secondary forest.The area-weighted average shape index decreased in grassland,shrub forest,and natural secondary forest.There was an increasing trend in plantation.The results showed that the grassland and natural secondary forests had increased fragmentation during 1989-2016,which would threaten the biodiversity and ecological security in the study area.(4)This study indicated that the optimal spatial grain size must be clearly defined first so as to elucidate the landscape pattern in a particular area.Combining the scale dependence characteristics of the landscape metrics and the information entropy calculation results,it can be seen that the effective grain range of scale dependence was 60-90m.After exceeding this scale dependence range,the landscape pattern information would lose,Finally,the study clarified that 90m was the suitable analysis grain size of vegetation landscape pattern in the study area.Thus,the characteristic scale was 90m.