CONTEXT

Rural communities in Guatemala and Colombia share a critical common challenge: the absence of accurate, up-to-date road data that limits access to services, hinders emergency response, and constrains territorial planning. In La Montaña de Jalapa, Guatemala, covering the municipalities of Jalapa and Mataquescuintla, dispersed rural communities depend on unpaved roads that deteriorate rapidly during rainy season, becoming impassable for extended periods and cutting off access to healthcare, education, and markets. The majority of these roads are absent from any official geographic information system, preventing adequate infrastructure investment and increasing community vulnerability to landslides, erosion, and flooding. Local authorities, disaster management agencies including CONRED, and development organizations identified this data gap as a key barrier to effective territorial planning. In the Colombian municipalities of Puerto Tejada and Padilla, in the department of Cauca, the challenge is compounded by historical underrepresentation in open cartographic databases. Puerto Tejada faces road blockages and deteriorating infrastructure linked to security dynamics, while Padilla has experienced recurring flooding and limited connectivity. In both municipalities, the absence of enriched road data, including surface type, width, and accessibility conditions, limits the capacity of local governments and communities to plan, respond to crises, and manage their territories effectively.

APPROACH

The project implements a replicable methodology combining AI-assisted spatial analysis, remote collaborative mapping, field data collection, and open data publication, adapted to the specific territorial context of each country.

In both Guatemala and Colombia, the process begins with a spatial analysis phase that integrates road data from OpenStreetMap with AI-detected road layers from Meta (MapWithAI) and Microsoft AI Roads. This analysis identifies potentially missing road segments, areas with poor connectivity, and roads lacking key attributes such as surface type and width. Results are structured as collaborative mapping tasks in HOT Tasking Manager and MapRoulette, enabling local volunteers, university students, and community mappers to validate and complete the road network using satellite imagery.

In Guatemala, field characterization is carried out by universtity students from Universidad Rafael Landívar and local officials using QField / Survey123 to document road dimensions, surface material, and accessibility conditions. Street-level imagery is captured using Mapillary, and a specialized technical team validated data quality before delivering three analytical cartographic products to the Presidential Office. CONRED participated actively, with local risk management officials trained to replicate the methodology in other at-risk areas.

In Colombia, field activities focus on the municipalities of Puerto Tejada and Padilla in Cauca. Drone imagery and 360° street-level photography complement remote mapping, with slope and flood risk indicators derived from digital elevation models to prioritize routes critical for evacuation, health access, and supply corridors. All data is published openly on OpenStreetMap, Mapillary, and OpenAerialMap.