The Study On The Dynamic Of Soil And Water Erosion In Nanfeng County Based On Landscape Pattern Change

For a long time,the shortage of water resources,the loss of land resources and pollution seriously threaten the sustainable development of social economy.Jiangxi is one of the three major rainstorm areas in the middle and lower reaches of the Yangtze River.Due to the specific geographical location and climate conditions,flood and drought disasters are frequent,and soil erosion is serious.Nanfeng County is a typical red soil hilly landform.At the end of last century and the beginning of this century,the industry of Nanfeng orange developed rapidly,and the planting area expanded rapidly,which led to the destruction of the original vegetation on the surface and exposed surface,which led to serious soil and water loss.On this basis,the landscape pattern optimization from the perspective of landscape pattern is aimed at alleviating the phenomenon of soil and water loss,and provides reasonable and scientific theoretical basis for soil and water conservation.This paper takes Nanfeng County as the research object,takes land use data,digital elevation model data,soil and water loss data as data sources,and comprehensively uses landscape ecology theory,BP neural network model and Arc GIS technology to analyze the impact of landscape pattern on Soil and water loss from the perspective of land use and terrain factors.Based on the above analysis,combined with the principal component analysis method,the BP neural network model was established to quantitatively describe the coupling relationship between landscape pattern and soil erosion and scenario simulation in the study area,and the landscape pattern optimization was carried out in order to reduce the soil and water loss.The results of this study can provide theoretical support and reference for the prevention and control of soil erosion in economic fruit tree planting area mainly based on citrus.The main results are as follows:(1)The landscape pattern and soil erosion change of Nanfeng County were analyzed by Arc GIS and FRAGSTATS software.The results showed that the landscape types of Nanfeng County were mainly forest landscape,cultivated land landscape and garden landscape,and forest land as the base landscape.From 1995 to 2015,the landscape pattern of Nanfeng County changed greatly,and the landscape of the garden mainly composed of Nanfeng orange increased in large areas,mainly distributed on both sides of Xujiang River and its tributaries with abundant water sources.The number of woodland and cultivated land is reduced.The types of soil and water loss in forest land are mainly mild,and the main types of soil and water loss in garden and cultivated land are very strong.From 1995 to 2015,the overall soil erosion in the study area fluctuates.(2)The correlation analysis of terrain factors and soil erosion.From 1995 to 2015,the landscape with slope of ? 5 ° has the largest soil erosion area,and the soil erosion area of the garden land on this slope has increased from 93.42km2 to 148.18km2.From the perspective of elevation,61.90% of the area in the study area is 100-249 m elevation zone,forest land landscape and garden landscape in 100-249 m elevation zone produce the most soil and water loss area,and the soil erosion rate of each landscape type in elevation zone ? 100 m is the highest.From 1995 to 2015,the soil and water loss rate of each elevation belt has an increasing trend.(3)The correlation analysis between landscape index and soil erosion data shows that the maximum patch index,landscape shape index,area weighted average shape index,edge density and other landscape indexes have significant correlation with soil erosion,but it is difficult to accurately describe the relationship between them with a single landscape index.(4)The BP neural network model between landscape pattern and soil erosion was established.Based on the discussion of the relationship between landscape elements such as land use,elevation,slope and soil erosion,it is found that there is a nonlinear mapping relationship between landscape index and soil erosion.Using the advantages of neural network to solve complex problems,six landscape index factors including landscape shape index,area weighted average shape index,similar adjacent proportion,patch cohesion index,aggregation index and normalized landscape shape index,were selected to establish the BP neural network prediction model of landscape pattern index and soil erosion in Nanfeng County In order to reduce the area of soil erosion and provide reference for the prevention and control of soil and water loss in southern red soil economic fruit tree planting area.