Mexico Road Network: AI-Derived Surface and Passability for 160,000 km

Maps 160,100 km of arterial roads in Mexico using AI-derived attributes from PlanetScope satellite imagery (2020 and 2024) and OpenStreetMap data. Produced by the Heidelberg Institute for Geoinformation Technology (HeiGIT), it provides surface type, width, and passability metrics for motorways, trunks, and primary/secondary roads. The dataset fills significant gaps where OpenStreetMap surface tags are missing.

Use Cases

• Evaluating logistical constraints for heavy transport using 'Road Width Class' and 'HPI' scores
• Identifying infrastructure improvements by tracking 'Road Surface Change' from 2020 to 2024
• Enhancing navigation datasets by filling missing 'surface' tags with 'DL Surface Prediction' data

Strengths

• 89.2% surface prediction accuracy
• Covers 160,100 km of the arterial road network
• Fills surface tag gaps for 38% of the OSM network in Mexico

Limitations

• Excludes minor and residential road classes
• AI-derived predictions may contain errors in specific environmental conditions despite high overall accuracy

Provenance

Source

HeiGIT (Heidelberg Institute for Geoinformation Technology)

Collection Method

Deep learning analysis of PlanetScope satellite imagery combined with OpenStreetMap vector data

Time Range

2020 - 2024

Freshness

Last updated March 2026; imagery analysis covers the period between 2020 and 2024.

Geography

Mexico