Phenotyping urban built and natural environments with high-resolution satellite images and unsupervised deep learning.

Cities in the developing world are expanding rapidly, and undergoing changes to their roads, buildings, vegetation, and other land use characteristics. Timely data are needed to ensure that urban change enhances health, wellbeing and sustainability. We present and evaluate a novel unsupervised deep clustering method to classify and characterise the complex and multidimensional built and natural environments of cities into interpretable clusters using high-resolution satellite images. We applied our approach to a high-resolution (0.3 m/pixel) satellite image of Accra, Ghana, one of the fastest growing cities in sub-Saharan Africa, and contextualised the results with demographic and environmental data that were not used for clustering. We show that clusters obtained solely from images capture distinct interpretable phenotypes of the urban natural (vegetation and water) and built (building count, size, density, and orientation; length and arrangement of roads) environment, and population, either as a unique defining characteristic (e.g., bodies of water or dense vegetation) or in combination (e.g., buildings surrounded by vegetation or sparsely populated areas intermixed with roads). Clusters that were based on a single defining characteristic were robust to the spatial scale of analysis and the choice of cluster number, whereas those based on a combination of characteristics changed based on scale and number of clusters. The results demonstrate that satellite data and unsupervised deep learning provide a cost-effective, interpretable and scalable approach for real-time tracking of sustainable urban development, especially where traditional environmental and demographic data are limited and infrequent.

Dhaka city water logging hazards: area identification and vulnerability assessment through GIS-remote sensing techniques.

Water logging is one of the most detrimental phenomena continuing to burden Dhaka dwellers. This study aims to spatio-temporarily identify the water logging hazard zones within Dhaka Metropolitan area and assess the extent of their water logging susceptibility based on informal settlements, built-up areas, and demographical characteristics. The study utilizes integrated geographic information system (GIS)-remote sensing (RS) methods, using the Normalized Difference Vegetation Water and Moisture Index, distance buffer zone from drainage streams, and built-up distributions to identify waterlogged zones with a temporal extent, incorporating social and infrastructural attributes to evaluate water logging effects. These indicators were integrated into an overlay GIS method to measure the vulnerability level across Dhaka city areas. The findings reveal that south and south-western parts of Dhaka were more susceptible to water logging hazards. Almost 35% of Dhaka belongs to the high/very highly vulnerable zone. Greater number of slum households were found within high to very high water logging vulnerable zones and approximately 70% of them are poorly structured. The built-up areas were observed to be increased toward the northern part of Dhaka and were exposed to severe water logging issues. The overall findings reveal the spatio-temporal distribution of the water logging vulnerabilities across the city as well as its impact on the social indicators. An integrated approach is necessary for future development plans to mitigate the risk of water logging.

Characterisation of urban environment and activity across space and time using street images and deep learning in Accra.

The urban environment influences human health, safety and wellbeing. Cities in Africa are growing faster than other regions but have limited data to guide urban planning and policies. Our aim was to use smart sensing and analytics to characterise the spatial patterns and temporal dynamics of features of the urban environment relevant for health, liveability, safety and sustainability. We collected a novel dataset of 2.1 million time-lapsed day and night images at 145 representative locations throughout the Metropolis of Accra, Ghana. We manually labelled a subset of 1,250 images for 20 contextually relevant objects and used transfer learning with data augmentation to retrain a convolutional neural network to detect them in the remaining images. We identified 23.5 million instances of these objects including 9.66 million instances of persons (41% of all objects), followed by cars (4.19 million, 18%), umbrellas (3.00 million, 13%), and informally operated minibuses known as tro tros (2.94 million, 13%). People, large vehicles and market-related objects were most common in the commercial core and densely populated informal neighbourhoods, while refuse and animals were most observed in the peripheries. The daily variability of objects was smallest in densely populated settlements and largest in the commercial centre. Our novel data and methodology shows that smart sensing and analytics can inform planning and policy decisions for making cities more liveable, equitable, sustainable and healthy.

Spatial heterogeneity in drinking water sources in the Greater Accra Metropolitan Area (GAMA), Ghana.

Universal access to safe drinking water is essential to population health and well-being, as recognized in the Sustainable Development Goals (SDG). To develop targeted policies which improve urban access to improved water and ensure equity, there is the need to understand the spatial heterogeneity in drinking water sources and the factors underlying these patterns. Using the Shannon Entropy Index and the Index of Concentration at the Extremes at the enumeration area level, we analyzed census data to examine the spatial heterogeneity in drinking water sources and neighborhood income in the Greater Accra Metropolitan Area (GAMA), the largest urban agglomeration in Ghana. GAMA has been a laboratory for studying urban growth, economic security, and other concomitant socio-environmental and demographic issues in the recent past. The current study adds to this literature by telling a different story about the spatial heterogeneity of GAMA's water landscape at the enumeration area level. The findings of the study reveal considerable geographical heterogeneity and inequality in drinking water sources not evidenced in previous studies. We conclude that heterogeneity is neither good nor bad in GAMA judging by the dominance of both piped water sources and sachet water (machine-sealed 500-ml plastic bag of drinking water). The lessons from this study can be used to inform the planning of appropriate localized solutions targeted at providing piped water sources in neighborhoods lacking these services and to monitor progress in achieving universal access to improved drinking water as recognized in the SDG 6 and improving population health and well-being.

Sachet water in Ghana: A spatiotemporal analysis of the recent upward trend in consumption and its relationship with changing household characteristics, 2010-2017.

The consumption of packaged water in Ghana has grown significantly in recent years. By 2017, "sachet water"-machine-sealed 500ml plastic bags of drinking water-was consumed by 33% of Ghanaian households. Reliance on sachet water has previously been associated with the urban poor, yet recent evidence suggests a customer base which crosses socioeconomic lines. Here, we conduct a repeated cross-sectional analysis of three nationally representative datasets to examine the changing demography of sachet water consumers between 2010 and 2017. Our results show that over the course of the study period sachet water has become a ubiquitous source of drinking water in Ghana, with relatively wealthy households notably increasing their consumption. In 2017, the majority of sachet water drinking households had access to another improved water source. The current rate and form of urbanisation, inadequate water governance, and an emphasis on cost recovery pose significant challenges for the expansion of the piped water supply network, leading us to conclude that sachet water will likely continue to be a prominent source of drinking water in Ghana for the foreseeable future. The main challenge for policymakers is to ensure that the growing sachet water market enhances rather than undermines Ghana's efforts towards achieving universal and equitable access to clean drinking water and sanitation.

Modeling the Impacts of Urban Flood Risk Management on Social Inequality.

The exposure of urban populations to flooding is highly heterogeneous, with the negative impacts of flooding experienced disproportionately by the poor. In developing countries experiencing rapid urbanization and population growth a key distinction in the urban landscape is between planned development and unplanned, informal development, which often occurs on marginal, flood-prone land. Flood risk management in the context of informality is challenging, and may exacerbate existing social inequalities and entrench poverty. Here, we adapt an existing socio-hydrological model of human-flood interactions to account for a stratified urban society consisting of planned and informal settlements. In the first instance, we use the model to construct four system archetypes based on idealized scenarios of risk reduction and disaster recovery. We then perform a sensitivity analysis to examine the relative importance of the differential values of vulnerability, risk-aversion, and flood awareness in determining the relationship between flood risk management and social inequality. The model results suggest that reducing the vulnerability of informal communities to flooding plays an important role in reducing social inequality and enabling sustainable economic growth, even when the exposure to the flood hazard remains high. Conversely, our model shows that increasing risk aversion may accelerate the decline of informal communities by suppressing economic growth. On this basis, we argue for urban flood risk management which is rooted in pro-poor urban governance and planning agendas which recognize the legitimacy and permanence of informal communities in cities.

A spatio-temporal land use and land cover reconstruction for India from 1960-2010.

In recent decades India has undergone substantial land use/land cover change as a result of population growth and economic development. Historical land use/land cover maps are necessary to quantify the impact of change at global and regional scales, improve predictions about the quantity and location of future change and support planning decisions. Here, a regional land use change model driven by district-level inventory data is used to generate an annual time series of high-resolution gridded land use/land cover maps for the Indian subcontinent between 1960-2010. The allocation procedure is based on statistical analysis of the relationship between contemporary land use/land cover and various spatially explicit covariates. A comparison of the simulated map for 1985 against remotely-sensed land use/land cover maps for 1985 and 2005 reveals considerable discrepancy between the simulated and remote sensing maps, much of which arises due to differences in the amount of land use/land cover change between the inventory data and the remote sensing maps.