South Africa Road Surface and Passability: 74,000 km of AI-Derived Data

HeiGIT produced this dataset covering approximately 74,000 km of arterial roads in South Africa using AI analysis of PlanetScope satellite imagery from 2020 and 2024. It provides deep-learning predictions for road surface, width, and a Humanitarian Passability Index (HPI) for segments classified as motorway, trunk, primary, and secondary in OpenStreetMap.

Use Cases

• Logistics planning using the Humanitarian Passability Index (HPI) to identify accessible routes for heavy trucks
• Infrastructure monitoring by analyzing Road Surface Change to detect unpaved-to-paved transitions over a four-year period
• Emergency response routing based on Road Width Class to determine vehicle size constraints for specific segments

Strengths

• 89.2% surface prediction accuracy compared to 64.7% for standard OSM tags
• Covers 74,000 km of the South African arterial road network
• Includes temporal change detection between 2020 and 2024 imagery
• Standardized Humanitarian Passability Index (HPI) for logistical planning

Limitations

• Restricted to major arterial road classes only
• AI-derived predictions may contain errors compared to ground-level surveys
• Geographic scope is limited to South Africa

Provenance

Source

HeiGIT (Heidelberg Institute for Geoinformation Technology)

Collection Method

AI-derived predictions from PlanetScope satellite imagery integrated with OpenStreetMap vector road segments.

Time Range

2020 - 2024

Freshness

The dataset uses satellite imagery from 2020 and 2024, with the most recent update in March 2026.

Geography

South Africa