Abstract. Road surface types not only influence the accessibility of road networks and socio-economic development but also serve as a critical data source for evaluating United Nations Sustainable Development Goal (SDG) 9.1. Existing research indicates that Africa generally have a low road paved rate, limiting local socio-economic development. Although the International Road Federation (IRF) provides statistical data on paved road length and road paved rates for certain African countries, this data neither covers all African country nor specifies the surface type of individual roads, making it challenging to offer decision-making support for improving Africa's road infrastructure. To fill this gap, this study developed the first dataset for 50 African countries and regions, incorporating the surface type of every road. This was achieved using multi-source geospatial data and a tabular deep learning model. The core methodology involved designing 16 proxy indicators across three dimensions—derived from five open geospatial datasets (OSM road data, GDP data, population distribution data, building height data, and land cover data)—to infer road surface types across Africa. Key findings include: The accuracy of the African road surface type dataset ranges from 77 % to 96 %, with F1 scores between 0.76 and 0.96. Total road length, paved road length, and road paved rates calculated from this dataset show high correlation (correlation coefficients: 0.69–0.94) with corresponding IRF statistics. Notably, the road paved rate also exhibits strong correlation with GNI per capita and HDI (correlation coefficients: 0.80–0.83), validating the reliability of the dataset. Spatial analysis of African road paved rates at national, provincial, and county scales revealed an average paved rate of only 17.4 % across the 50 countries and regions. A distinct "higher in the north and south, lower in the central region" pattern emerged, the average paved rate north of the Sahara is approximately three times that of Sub-Saharan (excluding South Africa). The African road surface type dataset developed in this study not only provides data support for enhancing road infrastructure and evaluating SDG 9.1 progress in Africa but may also facilitate research on how road surface types impact road safety, energy consumption, ecological environments, and socio-economic development.
Liu et al. (Wed,) studied this question.