Abstract
Population is a key indicator of socioeconomic development, urban planning and environmental protection, particularly for developing countries like China. But, census data for any given area are neither always available nor adequately reflect the internal differences of population. The authors tried to overcome this problem by spatializing the population across China through utilizing integer night-time imagery (Defense Meteorological Satellite Program/Operational Linescan System, DMSP/OLS) and land-use data. In creating the population linear regression model, night-time light intensity and lit areas, under different types of land use, were employed as predictor variables, and census data as dependent variables. To improve model performance, eight zones were created using night-time imagery clustering and shortest path algorithm. The population model is observed to have a coefficient of determination (R 2) ranging from 0.80 to 0.95 in the research area, which remained the same in different years. A comparison of the results of this study with those of other researchers shows that the spatialized population density map, prepared on the basis of night-time imagery, reflects the population distribution character more explicitly and in greater detail.
Acknowledgements
This study was supported by the Knowledge Innovation Project of the CAS, No. KZCX2-YW-Q03-07 and the National Science and Technology Supporting Projects of China, No. 2008BAH31B03. Dr Peng Gong, Dr Craig Cassells and the referees have provided plenty of suggestions to the article, as well as giving careful review and painstaking proofreading; we express our grateful appreciation here for their significant contribution to this article.