Many studies are pointing to the fact that cities are experiencing higher temperatures than non-built-up areas. Yet limited can be found on thermal inequalities in the context of vulnerable groups, specifically linked to people living in deprivation. Here, we study heat patterns across vulnerable groups living in deprivation as an important effort that should be paralleled to the other urban climate studies and aim at answering two primary questions: (1) how temperature varies within and across deprived areas, and (2) what the key driving factors are for such variation. We conduct intensive in-situ measurements by involving local residents in air temperature traverse across deprived neighbourhoods and modelling the pattern of air temperature with spatial covariates. We also compare different modelling techniques while securing the interpretability of the air temperature pattern by using understandable spatial covariates, which is especially informative for mitigation and adaptation, and linking scientific exploration and practical solutions.

Floods are one of the most destructive disasters worldwide and although they largely happen in rural, ruther than in urban areas, it is in the urban areas that substantial destruction of infrastructures is observed. Thus, cost effective methods to monitor flood damage and extent are required. In this paper, we investigate the implementation of U-Net on satellite and drone image dataset such as xBD and EDDA for building damage assessment in Mozambique. The recently published dataset EDDA was created by the National Institute for Disaster Management (INGD) and comprises drone imagery of Beira, in Mozambique. Using them, we obtained a dice score of 0.76 on building localization (BL) and mean intersection over the union (mIoU) of 0.54 on damage classification (DC). These are promising results considering that many datasets lack detailed information on African buildings. We also use some pre-trained models models such as ResNet for BL and DC.

Africa, as a major climate change hotspot, faces severe impacts, including extreme temperatures. Notably, urban areas are unequally affected by these impacts. The urban poor are particularly vulnerable to extreme temperatures, because of the environmental and physical characteristics of their neighbourhoods, and their limited resources to develop coping strategies. Limited knowledge exists of the spatial patterns of thermal inequalities within neighbourhoods. Our overall scientific objective is to explore the potential of Earth Observation (EO) to study how and why urban dwellers in the Global South (focusing on Africa) with different levels of deprivation are divergently exposed to varying temperatures and extreme heat, and to quantify the urban population exposed to such conditions. We make use of several state-of-the-art EO/AI models, and employ innovative in situ data collection methods together with local stakeholders through Citizen Science. We rely as far as possible on open or low-cost satellite imagery (e.g., Sentinel-1/2, Landsat, ECOSTRESS) for scalability and transferability, and we implement Machine Learning (ML) methods, including Deep Learning (DL). Results highlight significant local differences in thermal exposure, emphasizing the need to understand and communicate these spatial patterns to support the development of cost-effective adaptation strategies.

Mapping deprived urban areas in low- and middle-income countries is essential for policy development. While urban deprivation is a complex concept encompassing multiple dimensions, we propose an approach to capture its physical traits reflected in urban morphology, aiming for scalability. Our method makes use of affordable Earth Observation imagery and existing open geospatial datasets, and eliminates the need for manual labeling. It involves feature extraction, unsupervised learning, and pseudo-label based semi-supervised learning, resulting in 'soft' urban deprivation maps that avoid flagging areas as 'slums'. The study demonstrated its effectiveness in identifying the urban types associated with deprived areas at the scale of a large sub-Saharan African city.

Most research pertaining to mapping deprived urban areas is limited to locating and delineating deprived area's extents within and across cities. In this work, we go beyond and characterize deprived areas by utilizing a wide suit of remotely sensed predictors to map the intra-urban distribution of land cover (LC) in deprived communities in Nairobi, Kenya. We assess the contribution of WorldView-3 (WV-3) multispectral and shortwave infrared bands for the task of deprived urban areas land cover classification at a very-high-resolution scale. Our results highlight the potential of WV -3 to accurately map the LC while the potential of intra-urban transferability was shown to be satisfactory. Moreover, feature selection dramatically decreased the computational complexity of the LC models with no losses in classification accuracy. We propose a set of indicators such as the density of garbage piles to be extracted at an aggregated grid level. This aggregation helps characterize urban deprivation at a fine scale and assist local authorities and stakeholders in implementing evidence-based policy making.

A continuously growing number of urban inhabitants in Low- and Middle-income Country (LMIC) cities live in deprived areas. Such areas are under-serviced and characterized by poor living and environmental conditions, where the unplanned morphology interacts with physical hazards. While such areas proliferate, climate change has increasingly severe impacts on them. Deprived areas are often located in high-risk zones, e.g., flood zones. However, the absence of global databases on such areas is an obstacle to locate and prioritize hotspots of deprived communities exposed to climate change. Consequently, quantifying the numbers of exposed inhabitants is not possible, though it is required in support of the Sustainable Development Goals (SDGs) (e.g., 11, 13) and local adaptation strategies. Earth observation (EO) that allows producing such data fall short of providing city-level information due to computational constraints and unsolved methodological challenges related to scalability and transferability. This paper presents a framework to combine EO data with data on hazards (e.g., storms, floods) that impact urban areas and, in particular, deprived communities. First results in pilot cities show that deprived communities are systematically more exposed to physical hazards as compared to formal built-up areas. These problems are expected to intensify in the context of climate change, as most hazards will increase in their severity and frequency.

Earth Observation (EO) to produce policy-driven information on slums has been receiving increasing attention amongst experts. However, the geo-ethical concerns associated with making slum information publicly available are commonly neglected among the EO community. This study analysed the geo-ethics in terms of technology, product, and application-level using topic-focused interviews in the Greater Accra Region, Ghana. We identified that potential users have little knowledge of machine learning-based slum mapping methods, which implies the need for technology and product documentation to improve the acceptability and usability of EO data. We observed an application mismatch among institutions. While NGOs and research institutions required data for pro-poor initiatives, most government institutions needed data for slum eradication. Such mismatches require a rethinking of how slum data should be made public. We present a guide to disseminate information to users in support of developing a global slum data repository.

With rapid urbanization leading to the proliferation of deprived urban areas (often referred to as “slums”) in sub-Saharan Africa, there is a growing number of city dwellers living in inadequate housing conditions and being exposed to multiple hazards. In this context, Earth Observation has the potential for filling gaps in spatial data availability and thereby support evidence-based policy making. We assess the potential of free open-source software, open dual-pol SAR and optical imagery (Sentinel-l and Sentinel-2), and open global datasets for producing accurate city-scale maps of areas having morphological characteristics of deprivation. Implementing a grid-based machine learning approach, we evaluate different combinations of spectral and spatial Sentinel features, and features from global data. The results show that a high accuracy can be reached with the best combinations. Since publishing maps with hard labels (e.g., deprived vs. non-deprived areas) could raise ethical concerns or even lead to misuses, the output is provided as gridded morphological deprivation probability maps.

Landfill managers are subject to obligations which include the regular monitoring of the topographical and land cover (LC) evolution of the site. This research aims at developing a cost-effective non-intrusive methodology for mapping landfill LC features. To this end, a state-of-the-art OBIA open-source workflow based on an integration of GRASS GIS and Python programming environment was adapted and applied to 3-cm optical UAV image acquired over the landfill site of Hallembaye (Belgium). The results of this 8-class supervised classification are promising with an overall accuracy of 80.5%. This study shows that existing open-source processing chain can be adapted to UAV imagery. In addition, the added value of feature selection and of textural information provided by very high-resolution optical data is also highlighted. Finally, this study illustrates the potential of machine learning for the monitoring of landfill sites.

In this letter, the recently developed extreme gradient boosting (Xgboost) classifier is implemented in a very high resolution (VHR) object-based urban land use-land cover application. In detail, we investigated the sensitivity of Xgboost to various sample sizes, as well as to feature selection (FS) by applying a standard technique, correlation-based FS. We compared Xgboost with benchmark classifiers such as random forest (RF) and support vector machines (SVMs). The methods are applied to VHR imagery of two sub-Saharan cities of Dakar and Ouagadougou and the village of Vaihingen, Germany. The results demonstrate that Xgboost parameterized with a Bayesian procedure, systematically outperformed RF and SVM, mainly in larger sample sizes.