Rural green space is the foundation of rural environment. In order to solve the practical problems of rural green space, this paper starts from the perspective of rural space, and studies the systematic and normative extraction method of rural green space in Jiangning District of Nanjing city. It provides a scientific basis for the development of rural green space in China. Based on the object-oriented threshold classification method, this paper divides the rural green space into cultivated land, grassland, forest land, residential green space and road green space, and other features into water, road and urban and rural land, and analyzes the land coverage. The main research contents and conclusions are as follows: (1) The multi-scale segmentation parameters are determined. The influence of spectral factor, shape and compactness factor, segmentation scale and band weight on the experimental results is analyzed. After many experiments, the optimal segmentation scale parameters are determined as follows: water layer band weight 1:1:1:1:4, shape factor 0.1, compactness factor 0.5 and segmentation scale 60; The band weight of vegetation layer is 1:1:1:1:4, shape factor is 0.1, compactness factor is 0.5 and segmentation scale is 40; The band weight of urban and rural strata is 1:1:1:1:1, shape factor is 0.3, compactness factor is 0.5 and segmentation scale is 20. (2) The feature space rule sets of different land cover types are established. According to the spectral, geometric, texture and exponential characteristics of different land cover types, combined with sample analysis, the rule sets of extracting water body are determined as NDWI≥0.139, mean_ nir≤249; The rule set of green space and blue roof building is NDVI≥0.14; The rule set of blue roof building is NDSI≤- 0.19, mean_ Blue≥416; The rule set of road extraction is length ≥ 60, GLCM_ STdD≥44.5, BBI(374,453);The rule sets of grassland and cultivated land extraction were NDVI≤0.45, GLCM_ STdD≤36.1; The rule set of extracting cultivated land is rectangular fit≥0.74, NDVI≥0.16; The rule set of extracting road green space and residential green space is rel. Area of ≤20. (3) Object oriented threshold separation results and analysis. The results show that the overall accuracy of the method is 90.4%, and the kappa coefficient is 0.815. The total area of green space in the study area, including woodland, grassland, cultivated land, road green space and residential green space, is 1117413 m2, accounting for 87.7% of the total area. The area of other land types is 321148 m2, accounting for 12.3% of the total area. The rural greening rate in the study area is as high as 87.7%.
Read full abstract