Landscape Patterns Change Analysis In Yuanjiang Savanna Valley Based On Artificial Neural Network

Landscape pattern has always been the focus of landscape ecology research, the change of land use has a directly effect on landscape pattern of the earth surface. Yuanjiang savanna valley belongs to fragile ecology region, where the land use is irrational and make it has the potential problem of desertification. Based on theoretical basis of landscape ecology and artificial neural network, a part region of Yuanjiang savanna valley was taken as the study area, and two remote sensing images, which were got in 2006 and 2012, were interpreted with RS and GIS technology to get the land use data of the study area. Land use change model and land use type of transfer matrix analysis model was used to analyze the land use change regular in space and time scale of study area. Artificial neural network model is used to simulate landscape pattern dynamic change of study area and predict the change will happen in the future. The simulation study could quantitative analyze the change of landscape pattern in study area.Land use types in study area mainly contained cultivated land, woodland and savana. From 2006 to 2012, the areas of cultivated land, savana, tideland, residential land, garden plot and industrial warehouse space have increased, but the area of woodland and water area has reduced. The area change of cultivated land and woodland were very obvious, which were -106.70 km2 and 81.57 km2, respectively. Tideland had biggest dynamic index, which was 9.90%/a, and industrial warehouse space and water area also had a big dynamic index. Area increase of cultivated land was mainly from woodland and savanna and area increase of tideland was mainly from water area and savanna and area increase of industrial warehouse space was mainly from savana. Area reduce of woodland mainly became cultivated land and and garden plot increase, the landscape patch density will reduce. Landscape shannon’s diversity index will increase with the area proportion of residential land and garden plot increase, but will induce with the area proportion of cultivated land increase. Aggregation index will increase with the area proportion of water area, savanna and garden plot increase, but will reduce with the area proportion of tide land, residential land and cultivated land increase. With the area proportion of woodland, industrial warehouse space and tideland increase, landscape patch shape will become more complex, but with the area proportion of water area, savanna and garden plot increase, patch shape will become standardization.