Data analytics and sustainable urban development in global cities

Urban mobility and amenities

A number of our studies focus on accessibility to amenities and the impact of amenities on urban life in developing cities. Amenities, ranging from retail to parks to culture, have long been feted as a key driver of urban growth, with strong evidence for US cities (Glaeser et al., 2001). Evidence for the impact of amenities in developing cities is more limited, with limited signals of a strong impact on urban growth in Colombia (Duranton, 2016). In this special issue we have three studies focused on amenities and mobility in Latin American cities.

While there has traditionally been a lack of data on amenities in developing cities, recent increased availability of large datasets of amenities derived from crowdsourced map data (e.g., Google Places) has enabled researchers to explore these questions in more detail. In this SI, Heroy et al. (2023) deploy Google Places data to investigate the impact of the density of local amenities on urban mobility and travel modes in Bogota, Colombia. A key tenet of urban planning is that a high density of local amenities is expected to induce residents to shop and engage in leisure activities locally, thus reducing their travel radius and duration and presumably enhancing their quality of life. Combining the amenity data with travel surveys, Heroy et al. (2023) find that wealthy groups have a higher propensity to walk and shorter driving time if they live in an area of high amenity density, but there is no statistical effect for mid and low-income groups.

Graells-Garrido et al. (2023) investigate the interplay between physical (or material) and digital activities in urban ‘areas of interest’. Combining data on visits to both locations and websites derived from mobile phone data with crowdsourced amenity data from OpenStreetMap (an open access analogue of Google Places, albeit typically less complete), the authors develop a new method to identify areas of interest, and apply their method to Santiago, Chile. Their results indicate that several types of area of interest exist, including those dominated by physical amenities, those with a high diversity of both physical amenities and digital activities, and those with mainly digital activities which are located mainly in the periphery of the city.

Tomasiello et al. (2023) explore accessibility to leisure amenities such as green space and theatres in Sao Paulo, Brazil, focussing on differences between socio-economic and racial groups. The authors distinguish between public and private transport, and control for traffic conditions during the day using tracking data. They find that accessibility to leisure is higher for private transport users, decreases from the central to the peripheral areas, and is lowest for poor black residents.

Finally, Neto-Bradley et al. (2023) focus on the impact of access to clean cooking fuel in the residential energy transition. They develop a spatial microsimulation model which accounts for access to clean fuels – as well as socio-economic variables and local cooking practices – to predict cooking fuel consumption, and apply their model to four cities in the South Indian states of Kerala and Tamil Nadu for which primary data is available for comparison.

COVID-19 mobility

There has been a deluge of research on the effects of the COVID-19 pandemic, with a large number of studies investigating the impact of COVID-19 restrictions on mobility, both at the within city scale (e.g. Heroy et al., 2021) and the country scale (e.g. Lee et al., 2021; Xiong et al., 2020). Many of these studies have harnessed mobile phone data and other forms of ‘big’ mobility data. In this SI, we expand this literature in two directions: the first focused on public transport use in Bogota, Colombia, and the second on return migration of rural workers to Chinese cities.

Garcia-Arteaga et al. (2023) investigate the impact of the COVID-19 pandemic on transit patterns for the TransMilenio in Bogotá, perhaps the world’s most well-known Bus Rapid Transit system that is the main mode of transport for 10 million urban residents. Using tap-in data on users’ entry points into the system, the authors deploy a massive dataset on Transmilenio usage to construct a sequence of mobility networks representing inter-station visits before and during the pandemic. Analysing these networks, they detect a clear shift in station centrality or importance from north to south as the pandemic restrictions hit, reflecting the travel patterns of essential workers.

Liu et al. (2023) work at a larger scale, focussing on the return to work of China’s migrant workers during the early pandemic period. The authors deploy mobile phone data and machine learning to detect 167m rural migrant workers in 2020, focussing on three large urban agglomerations: the Beijing–Tianjin–Hebei Region, Yangtze River Delta and Pearl River Delta. They found that COVID-19 caused considerable disruption to typical migrant work patterns, both stranding workers out of place in some places, and causing worker shortages in others.

Urban form and resilience

A final cohort of articles harness machine learning and urban simulation models to investigate and predict urban form, growth and resilience. There is an expansive literature on urban growth models aiming to predict the expansive and use of urban land and/or population. However, a lack of data and a poor understanding of informality, from the informal economy to informal housing, has hindered the application and adoption of these models in developing cities. In this SI, a variety of studies aim to adapt these models to new landscapes.

At the street scale, De Carvalho Filho et al. (2023) investigate the characteristics of ‘void street interfaces’ (VSI), which are building frontages with no public function, limited ground-floor accessibility and restricted visual interaction, and share characteristics with inaccessible and opaquely fronted areas such as those used for gated communities. Concerned by the risks of segregation and exclusion posed by VSI, the authors construct a predictive machine learning algorithm to identify VSI based on socio-economic data in the census using training data from Recife in Brazil. They then apply the model to identify VSI across a wider set of Brazilian cities, observing, for example, that VSIs tend to locate on local rather than main roads.

Focussing on large scale urban form and function, Agyemang et al. (2023) develop an agent-based urban growth model-which describes the geospatial behaviour of key urban development actors such as households, real estate developers and government-to predict the location, legal and socio-economic status of future residential developments. A key novelty of the study, informal developments in the model are those that take place outside of residential zoned land, or are unplanned or untitled. The model is calibrated using local data for Accra, Ghana, and shown to outperform alternative models such as SLEUTH, particularly when it comes to informal characterization.

Interested in urban planning and resilience, Guan et al. (2023) focus on the protection of urban water sources such as rivers. Deploying urban growth models to predict population and land use growth under various scenarios, the authors use the city of Jiaxing, China as a case study to investigate the optimal width of water protection zones around water sources. The authors find that a protection zone of about 400 m represents a good balance between water protection and avoiding overly fragmented urban areas.

Conclusions

At least two observations can be made on the basis of the articles brought together in this special issue. First, it is evident that data analytics are being used to address a wide range of topical concerns in rapidly growing cities beyond the global North. These issues both mimic those addressed in data analytical research focused on Northern cities (e.g. access to green space and other local amenities) and also differ from them (e.g. access to clean cooking fuels). In both instances the approaches and methods used are mostly similar to those in research on cities in the global North. This is perhaps to be expected in light of the epistemological premises underpinning Urban Science, but also offers an interesting contrast with lively debates in urban studies in disciplines like human geography, sociology and anthropology. There, a contentious and unresolved question keeps unfolding about whether cities in the global South and East require different theories from those in Western Europe, the USA, Canada, etc. More specifically, the debate is whether empirical differences between cities in different parts of the world reflect qualitative differences in the constitutive processes and logics that underpin and shape urban development or not (e.g. Roy, 2016; Randolph and Storper, 2023). Work on data analytics and Urban Science could meaningfully contribute to this debate, and the evidence brought together in compendiums such as this special issue can be drawn upon to do so.

Second, the articles included in the special issue, and indeed the set of abstracts that were submitted originally, remain geographically uneven. They are disproportionally focused on cities in middle income countries, in part because more (digital) data are available on those cities than in their counterparts in low-income countries. A related unevenness is that researchers based in or from global North countries are clearly overrepresented among the total set of authors included in this special issue. Perhaps expectedly, the rise of data analytics seems to be reinforcing historically emerged geographical inequalities and inequities in the production of advanced quantitative knowledge, and this is an area of concern-at least to us. While the work brought together in this special issue is very welcome, we also believe it confirms and reinforces the need for greater local capacity building regarding data analytics in academic and other knowledge creating institutions across the world’s low-income countries.