Improved GDP spatialization approach by combining land-use data and night-time light data: a case study in China’s continental coastal area

<p>Gross domestic product (GDP) reflects a nation or region’s economic growth as a whole, and is the sum of product in the primary, secondary, and tertiary sectors of the economy in the area. However, statistical GDP data is problematic in integrated application with geographical data. The GDP spatialization data, which shows the GDP in grid cells and often is obtained by operating a spatialization model, is more useful than its officially published statistical data recorded by administrative units in both spatial representation and application. Thus, there is a need to improve the GDP spatialization models, and to present these models in a way as clear and transparent as possible. In this article, by taking China’s continental coastal area as a case study area, we combined economic census data, land-use data, and night-time light data together, and developed a technique that we call the ‘dynamic regionalization’ method to improve the GDP spatialization products. We then created GDP spatialization models for three sectors of the economy (i.e. the primary, the secondary, and the tertiary sector) in 2000, 2005, and 2010, respectively. We find the following. (1) Because the ‘overglow’ effect of night-time light data has a bad influence on spatialization models, we used land-use data to distinguish the distribution plots of the tertiary sector on night-time light images. Compared with setting a threshold merely, land-use data can more effectively remove the ‘overglow’ effect. (2) Owing to the prominent spatial heterogeneity of GDP distribution in China’s continental coastal area, building one spatialization model for the whole area would probably produce the estimated products with poor accuracy, so the ‘dynamic regionalization’ method was adopted to dynamically divide the whole study area into several subregions, and build separate spatialization models for each subregion. The accuracy assessment showed that the new method improved the accuracy of GDP spatialization data, especially in the area with high spatial heterogeneity.</p>