Satellites and machine learning are mapping Durban's landslide risks before the next storm

In April 2022, floods and landslides tore through eThekwini — the municipality that includes Durban, South Africa — displacing more than 40,000 residents. A new study in Scientific Reports aims to make sure the city knows its danger zones in advance. Cher Petersen and colleagues combined satellite imagery with machine learning to map exactly where the land is most likely to slide, and found that road building, property development and vegetation loss are sharply raising the risk in places. A map cannot stop the rain. But a city that knows which hillsides to watch, drain and leave undeveloped is a city that can act before the emergency, not after it.

SDG 11 · Sustainable cities and communities — Mechanism: tech innovation