Identifying Urban Areas by Combining Data from the Ground and from Outer Space : An Application to India

Galdo, Virgilio ; Li, Yue ; Rama, Martin

World Bank, Washington, DC

关键词: URBANIZATION;    URBAN EXTENT;    URBAN SPRAWL;    SATELLITE IMAGERY;    GEOSPATIAL ECONOMICS;   

DOI  :  10.1596/1813-9450-8628
RP-ID  :  WPS8628

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

This paper develops a tractable methodto identify urban areas and applies it to India, whereurbanization is messy. Google Earth images are assessedsubjectively to determine whether a stratified large sampleof Indian cities, towns and villages, as officially defined,are urban or rural in practice. Based on these assessments,a regression analysis combines two sources ofinformation—data from georeferenced population censuses anddata from satellite imagery—to identify the correlates ofunits in the sample being urban. The resulting model is usedto predict whether the other units in the country are urbanor rural in practice. Contrary to frequent claims, India isnot substantially more urban than implied by census data.And the speed of urbanization is only marginally higher thanofficial statistics suggest. But a considerable number oflocations are misclassified in the midrange between villagesand state capitals. The results confirm the value ofcombining subjective assessments with data from thesedifferent sources.