Integrating Activity-Based Geographic Information and Long-Term Remote Sensing to Characterize Urban Land Use Change

Cheng Fu,Kathleen Stewart,Xiao-Peng Song

doi:10.3390/rs11242965

Cheng Fu, Kathleen Stewart + Show 1 more

Open Access

https://doi.org/10.3390/rs11242965

Copy DOI

Abstract

The land use structure is a key component to understand the complexity of urban systems because it provides a snapshot of urban dynamics and how people use space. This paper integrates socially sensed activity data with a remotely sensed land cover product in order to infer urban land use and its changes over time. We conducted a case study in the Washington D.C.–Baltimore metropolitan area to identify the pattern of land use change from undeveloped to developed land, including residential and non-residential uses for a period covering 1986–2008. The proposed approach modeled physical and behavioral features of land parcels from a satellite-based impervious surface cover change product and georeferenced Tweets, respectively. A model assessment with random forests classifiers showed that the proposed classification workflow could classify residential and non-residential land uses at an accuracy of 81%, 4% better than modeling the same land uses from physical features alone. Using the timestamps of the impervious surface cover change product, the study also reconstructed the timeline of the identified land uses. The results indicated that the proposed approach was capable of mapping detailed land use and change in an urban region, and represents a new and viable way forward for urban land use surveying that could be especially useful for surveying and tracking changes in cities where traditional approaches and mapping products (i.e., from remote sensing products) may have a limited capacity to capture change.

Highlights

For each ISC object, a set of physical properties were derived from pixels in the Impervious Surface Cover Annual Change Map (ISC-ACM) data as the physical signatures, and a set of activity properties were derived from associated Tweets as the activity signatures; together, these were combined to form the feature set of the ISC object
The output of the framework maps the spatial details of land use change and profiles the trajectories of different land use types over time, which contributes to an understanding of the evolution of urban development as a complex system
For municipalities or counties in the US with zoning or land use maps, the output of this framework may help to address mapping errors in current maps. This framework utilizes remote sensing imagery to model the physical signatures of land cover and georeferenced Tweets to model activity signatures associated with different land use types

Summary

Introduction

It is predicted that the global urban area may triple, from the year 2000, by 2030 [2]. Information on land use (LU, the social function of land) is important to understand the dynamics and complexity of urban systems. The intra-city land use structure can benefit models of carbon emission estimations [3,4], hazard resilience [5], and transportation [6,7]. Information on the extent of urban sprawl (i.e., uncoordinated city growth [8]) is often missing or outdated. Official zoning maps or land use maps based on land surveying are often not updated frequently, due to financial and time costs, and do not capture the rapid land changes accompanying urbanization

Objectives

Methods

Results

Discussion

Conclusion