Classification of cities in Bangladesh based on remote sensing derived spatial characteristics

Abstract

In Bangladesh, cities are conventionally classified based on population size and revenue collection. This conventional city classification system neglects the spatial characteristics inherit in cities. Providing a more comprehensive city classification system is essential for the country's future budget allocation and infrastructure development. Five spatial features: city size (area), urban form (AWMPFD), the ratio of built-up and non-built-up areas, urban growth rate, and total night lights intensity for 331 cities of Bangladesh are derived from remote sensing data. This study classifies these cities into six classes using a hierarchical clustering algorithm based on five selected spatial characteristics. The six categories are named for their levels of spatial development, with Cluster 1 being the highest level and Cluster 6 being the lowest level. The share of employment in the primary sector (agriculture) gradually rises from Cluster 1 to Cluster 6. In contrast, the employment share of the service sector follows a reverse trend from Cluster 2 to Cluster 6. Both per capita income and expenditure is higher for the large cities of Cluster 2 than for the metropolitans of Cluster 1. Comparisons across the six classes with non-spatial attributes validate the classification system. Findings also reveal that remote sensing derived spatial information can explain non-spatial characteristics of cities. Therefore, remote sensing derived spatial attributes of cities can be used for city classification where census data are scarce.