Urban Expansion In China Over The Past30Years Detected Using DMSP/OLS Nighttime Light Data

Urban expansion has an important influence on land use and cover change as well as the sustainable development of urban area. During the past30years, China has experienced rapid development in urban areas, but there is also the problem of out-of-control space expansion. Extracting urban information with remote sensing data can quickly reveal the urban expansion speed, expansion mode and so on. The urban expansion process can be overall grasped through change analysis. Nighttime light imagery of the U.S. Air Force Defense Meteorological Satellite Program (DMSP) Operational Linescan System (OLS) has been proved to be a valid data source to monitor human activities and urban research.Taking the Administrative Region of China as the study area, the main purpose of this research is to investigate the urban expansion for the past30years with DMSP/OLS nighttime light data. Firstly, the study presented the image representation of the urban area based on the nighttime light data. Secondly, intercalibration is used to improve comparability of nighttime light data form1992-2010. Then, morphology and contour information was extracted using Support Vector Machine (SVM) classification methods in partition. Lastly, urban expansion of China in long time series was analyzed. The main contents and results of the research are presented as follows:(1) Urban characterization on the nighttime light image was analyzed. The feature representing the morphology and contour information of urban was investigated on the DMSP/OLS imagery. There is an obvious "overflow" phenomenon between the block or patchy distribution of urban area in the nighttime light data and the reference urban area. Silhouette, shape and structure information of urban can be obtained in the nighttime light image by adjusting brightness value from low to high. Road information between urban shows obvious line distribution on the nighttime lights imagery, but it’s hard to express correctly. It was because the brightness value was too small and could be easily confused with the surrounding overflow pixels of urban.(2) The multi-temporal nighttime light data was intercalibrated. The DMSP/OLS data has no on-board calibration, so it’s lack comparability between data in different years. This paper determines the correction parameters of19years from1992-2010average nighttime light radiation calibration data on the basis of existing research methods. Then the intercalibration was carried out for the multi-temporal nighttime light imagery.(3) A partition SVM-based method was proposed to extract the shape and contour information of urban area. The study area was divided into7areas, taking the effects of unbalanced economic development in China into account. Extracting urban information with the SVM classifier in partition areas, we can get better spatial shape and more stable accuracy of the urban.(4) Urban expansion of China and typical urban agglomerations was analyzed for the past30years. Urban area in the recently30years of China showed exponential growing pattern from the global perspective, while it was quite different in different regions. For the three typical urban agglomerations (Yangtze River Delta, Pearl River Delta, Jingjinji Region), the general trend of expansion intensity can be summarized as "increase-decrease-and then increase" form and the gathering form or agglomerations trend gradually later. The Yangtze River Delta city group has the obvious transfer trend of gravity center, which was kept moving to the northwest in the past30years. In addition, the three major agglomerations had been in the most active area of China’s urban development from the trend of relative changes.As a whole, a technical idea of extraction and expansion of urban information was established with DMSP/OLS nighttime light data considering the characteristics of urban development in recently30years. Nighttime light image has the ability to extract the shape and contour information of urban through investigating the characterization of urban information on the nighttime light imagery. Secondly, better extraction of spatial shape and more stable accuracy can be achieved using the partition SVM classification method. The spatial pattern of urban expansion was constructed based on the extraction results of1981-2010years. The study provides practical technical route and data support for urban expansion.