Evaluating the Ability to Use Contextual Features Derived from Multi-Scale Satellite Imagery to Map Spatial Patterns of Urban Attributes and Population Distributions

Steven Chao,Michael Mann,Adane Bedada,Ryan Engstrom

doi:10.3390/rs13193962

Abstract

With an increasing global population, accurate and timely population counts are essential for urban planning and disaster management. Previous research using contextual features, using mainly very-high-spatial-resolution imagery (<2 m spatial resolution) at subnational to city scales, has found strong correlations with population and poverty. Contextual features can be defined as the statistical quantification of edge patterns, pixel groups, gaps, textures, and the raw spectral signatures calculated over groups of pixels or neighborhoods. While they correlated with population and poverty, which components of the human-modified landscape were captured by the contextual features have not been investigated. Additionally, previous research has focused on more costly, less frequently acquired very-high-spatial-resolution imagery. Therefore, contextual features from both very-high-spatial-resolution imagery and lower-spatial-resolution Sentinel-2 (10 m pixels) imagery in Sri Lanka, Belize, and Accra, Ghana were calculated, and those outputs were correlated with OpenStreetMap building and road metrics. These relationships were compared to determine what components of the human-modified landscape the features capture, and how spatial resolution and location impact the predictive power of these relationships. The results suggest that contextual features can map urban attributes well, with out-of-sample R2 values up to 93%. Moreover, the degradation of spatial resolution did not significantly reduce the results, and for some urban attributes, the results actually improved. Based on these results, the ability of the lower resolution Sentinel-2 data to predict the population density of the smallest census units available was then assessed. The findings indicate that Sentinel-2 contextual features explained up to 84% of the out-of-sample variation for population density.

Highlights

The world population is projected to reach 9.8 billion by 2050, with most of the growth in developing countries and with 68% predicted to live in urban areas [1]
192 Gram Niladhari Divisions (GN)) were used to model urban attributes to investigate what contextual features derived from VHSR imagery represent in the human-modified landscape
This study analyzed the ability of contextual features to model attributes of the humanmodified landscape and population density

Summary

Introduction

The world population is projected to reach 9.8 billion by 2050, with most of the growth in developing countries and with 68% predicted to live in urban areas [1]. While useful, has a number of limitations: (1) countries usually conduct censuses at most once every 10 years, as recommended by the United Nations [13], which affects their relevance, as high migration and urban growth rates can make existing data quickly outdated [19]; (2) due to privacy reasons, census data are usually aggregated, generalized, and not available at the local scale [20,21,22]; (3) census units do not necessarily align with human settlement boundaries [20,21,22,23]; (4) censuses are resource intensive, which. Remote sensing technologies enable frequent data collection, making them effective and widely used in predicting population [5,26]. Wu et al, [27]

Methods

Results

Discussion

Conclusion