| Impervious surface areas (ISA) are anthropogenic features through which water cannot infiltrate into the soil, such as roads, square, rooftops, and so on. As a kind of typical components of the surface coverage, impervious surface is regarded not only as an indicator of the degree of urbanization, but also a major indicator of environmental quality. In this paper, with Yangtze River Delta (YRD) of high urbanization level as research region, Landsat OLI and MODIS as key data, we extracted ISA through Decision Tree (DT), Linear Spectral Mixing Analysis (LSMA) and Complement of Vegetation Fraction (CVF), respectively. On this basis, the spatial distribution, heterogeneity and driving force analysis of ISA in YRD were studied. In addition, we investigated the influence and uncertainty induced by multiple data sources and modelling methods.(1) The extracting accuracies of ISA through Landsat OLI by means of DT and LSMA were 0.68 and 0.73. Similar extracting accuracies through MODIS data by means of DT and CVF were 0.62 and 0.60. Regions with higher extraction accuracy were mainly in the belt through Shanghai-Suzhou, Wuxi, Changzhou-Nanjing areas and some cities in north Zhejiang Province, such as Hangzhou, Jiaxing and so on.(2) The coverage rate of ISA in YRD was 10.31%, with obvious spatial heterogeneity. The coverage rate of ISA, with the belt of Shanghai-Suzhou, Wuxi, Changzhou-Nanjing as central axis, decreased gradually to both sides, And the spatial characteristics of ISA in Jiangsu Province, Zhejiang Province and Shanghai were exhibited as stars distribution, zonal distribution and concentric distribution as a whole, respectively.(3) The coverage rate of ISA in YRD was influenced by comprehensive factors as economy, population, position condition and natural topography. ①. The coverage rate of ISA showed significant positive correlation with gross domestic product (GDP) and GDP per person (correlation coefficient[R] were 0.88 and 0.81, p< 0.01, respectively), which indicated that economic development plays an important role in the expansion of ISA.②.The coverage rate of ISA showed significant linear positive correlation with population (R=0.67, p<0.05), the change of population had a direct impact on urban land use pattern. ③.The coverage rate of ISA showed significant negative correlation with the distance from the city to Shanghai (R=-0.67, p<0.01), which showed that as the center city of YRD, Shanghai had important radiation effect on the surrounding cites. ④.The coverage rate of ISA in YRD reached its max value (17.03%) in the elevation interval of 5-10 m. The area and its coverage rate of ISA decreased step by step when the elevation over 10 m. Apart from this, government policy also had an influence on the spatial distribution of ISA.(4) The proportion of same pixel from all 4 results was 74.71% in which pixel value of 0 (i.e., pervious surface) was 63.18% and pixel value of 4 (i.e., ISA from four methods) was 11.53% in the total uncertainty evaluation in the extraction of ISA in YRD. Further studies showed that uncertainty within the extraction of ISA through Landsat OLI and MODIS by means of DT mainly resulted from the difference between image quality, sensor spectral and the loss of precision when mosaicking and resampling; Uncertainty of extraction of ISA through Landsat OLI by means of DT and LSMA indicated the result of LSMA was more precise than DT, and DT had no access to the difference and connection of spectrum curve between mixed pixel and pure pixel; Uncertainty of extraction of ISA through MODIS by means of DT and CVF indicated mixed pixels including ISA and vegetation were easy to be extracted as ISA through CVF with coarse spatial resolution. |
PDF Full Text Request