Using machine learning to build safer cities
In eThekwini, researchers have harnessed machine learning and satellite data to spot the city's most landslide-prone zones. A new study published in Scientific Reports has found that urban development activities, including road construction, property development and vegetation loss, are significantly increasing landslide susceptibility across parts of eThekwini.
Researchers in South Africa have utilized machine learning and satellite imagery to identify the most landslide-prone zones within the city of eThekwini. This innovative study, published in Scientific Reports, reveals that urban expansion activities, such as road construction and property development, are significantly raising the risk of landslides in certain parts of the city.
The findings come just four years after a devastating event in April 2022, which caused hundreds of fatalities, the displacement of over 40,000 residents, and an estimated R36 billion worth of damage to infrastructure and businesses.
The study employed advanced machine learning models to map areas with the highest likelihood of landslides, taking into account factors like land use, elevation, vegetation cover, and geological conditions. The research pinpointed that areas with built-up development and cultivated land are more susceptible to landslides compared to less disturbed environments, as removing vegetation can compromise slope stability and construction activities can disrupt drainage patterns, adding pressure to vulnerable landscapes.
Dr. Laven Naidoo, a senior researcher from the Gauteng City-Region Observatory, emphasized the importance of understanding these risk factors to inform preventive planning and risk reduction in the face of increasingly frequent extreme weather events. The study used two machine learning models, random forest and support vector machines, which achieved accuracy levels above 98% in identifying potential landslide hotspots.
The results indicated that coastal areas of eThekwini are particularly vulnerable to landslides, while mountainous inland regions are generally at lower risk. This information can guide municipalities and planners in managing urban development more effectively and prioritizing disaster preparedness efforts. Traditional landslide assessments are often labor-intensive and time-consuming, but the combination of machine learning and satellite data offers a more efficient and accurate approach to identifying at-risk areas.
The study builds on lessons learned from the April 2022 floods, which were one of the most destructive natural disasters in KwaZulu-Natal's history. The floods affected both informal and formal settlements, highlighting the broader implications of landslide risk beyond traditionally vulnerable communities. The researchers noted that while poor households in informal settlements face higher exposure to environmental hazards, formal residential developments were also impacted, raising questions about development practices, vegetation removal, and the suitability of building on geologically sensitive terrain.
Overall, the findings underscore the complex interplay between environmental factors and human decisions in landslide occurrence. As climate change increases the likelihood of extreme rainfall events, the use of advanced technologies like machine learning, remote sensing, and satellite monitoring becomes crucial for providing valuable decision-support tools, especially in resource-constrained contexts.
Written by urgent.news from Phys.org's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
