Abstract: Gross Domestic Product (GDP), which provides the characteristics of social and economic development, is the most widely used socioeconomic indicator for the politic agency to develop strategies and assess disaster risks. However, there often exists a spatial mismatch between socio-economic data (at administrative unit level) and environmental data (at pixel level). In this paper, a Human Settlement Index (HSI) was developed by combining the Defense Meteorological Satellite Program Operational Linescan System (DMSP/OLS) nighttime light imagery data and enhanced vegetation index (EVI) data. The input of EVI can effectively decrease the information loss of population distribution caused by the saturation effect of nighttime light imagery. A threshold method (Digital Number<9) was used to remove the potential nighttime light blooming areas to reduce the error caused by the overglow effect of DMSP/OLS nighttime light imagery and classify the different types of GDP. The land use and land cover data was used to map the agricultural GDP, and the non-agricultural GDP was mapped based on the linear relationship between the cumulative HSI and the GDP of secondary and tertiary industries (R2=0.823). Then the GDP density across the coastal area of China in 2010 was mapped at a spatial resolution of 250 m. The results showed that the application of HSI can improve the accuracy in simulating the GDP of secondary and tertiary industries than the single nighttime light model, and the highest GDP density mainly located in the regions with a high urbanization level, especially in the Yangtze River Delta and the Pearl River Delta. The method developed by this paper to map GDP at pixel level can compare the GDP value within and between the study units. This method can provide an effective application in developing economic strategies, regional planning, and natural disaster risk assessment, etc.
Key words：GDP spatialization; DMSP/OLS; EVI; land use; coastal area