Indonesia Road Surface and Passability: 121,400 km of AI-Derived Metrics

Delivering AI-derived road surface, width, and passability metrics for approximately 121,400 km of arterial roads in Indonesia, produced by HeiGIT using 2020 and 2024 PlanetScope satellite imagery. It covers motorway, trunk, primary, and secondary road classes, filling data gaps for the 66% of Indonesian OpenStreetMap segments that lack surface tags. The analysis utilizes a deep-learning model with 89.2% accuracy to classify infrastructure across the archipelago.

Use Cases

• Assessing logistical accessibility for disaster relief using the Humanitarian Passability Index (HPI) scores
• Identifying infrastructure development trends by filtering for 'unpaved to paved' transitions in the Road Surface Change column
• Route planning for heavy logistics by selecting segments with Road Width Class 3 (width >5.5 m)

Strengths

• Surface model achieves 89.2% accuracy compared to 64.7% for standard OSM surface tags
• Provides consistent coverage across 121,400 km of the Indonesian arterial road network
• Includes temporal change detection for infrastructure development monitoring between 2020 and 2024

Limitations

• Restricted to arterial road classes (motorway, trunk, primary, secondary), excluding local and residential streets
• Humanitarian Passability Index is a modeled estimate rather than real-time ground-verified status
• AI-derived predictions may contain errors in areas with dense tropical canopy or persistent cloud cover

Provenance

Source

Heidelberg Institute for Geoinformation Technology (HeiGIT)

Collection Method

Deep-learning predictions derived from PlanetScope satellite imagery (3m resolution) integrated with OpenStreetMap road geometries.

Time Range

2020-2024

Freshness

Based on satellite imagery from 2020 and 2024; last updated March 2026.

Geography

Indonesia