GDP Spatialization in City of Zhengzhou Based on NPP/VIIRS Night-time Light and Socioeconomic Statistical Data Using Machine Learning

Abstract

This article introduces a comprehensive methodology for mapping and assessing the urban built-up areas and establishing a spatial gross domestic product (GDP) model for Zhengzhou using night-time light (NTL) data, alongside socioeconomic statistical data from 2012 to 2017. Two supervised sorting algorithms, namely the support vector machine (SVM) algorithm and the deep learning (DL) algorithm, which includes the U-Net and fully convolutional neural (FCN) network models, are proposed for urban built-up area identification and image classification. Comparisons with Municipal Bureau of Statistics data highlight the U-Net neural network model exhibits superior accuracy, especially in areas with diverse characteristics. For each year from 2012 to 2017, a spatial GDP model was developed based on Zhengzhou's urban GDP and U-Net sorted images. This research provides valuable insights into urban development and economic assessment for the city.