Researchers have developed a methodology that uses remotely sensed data measurements to predict completeness in the coverage of building footprint data in OpenStreetMap. The methodology, used to analyze three islands, Haiti, Dominica, and St. Lucia, was able to explain up to 94% of the variation in building footprint coverage. Being able to identify areas with poor GIS data coverage can be used to help prioritize mapping efforts, especially for use in disaster risk management.
OpenStreetMap is a crowdsourced project that uses volunteers to develop and edit geospatial datasets covering the world’s streets, buildings, parks, and other features. Coverage and accuracy of the collected datasets varies between and within countries. Using remotely sensed data from satellite sources, researchers developed a methodology to measure the completeness of OpenStreetMap data for building footprints by identifying “mapping gaps”, areas where footprints have not yet been mapped. To do this, predictors based on remotely sensed data were used to estimate the presence of OpenStreetMap building footprints. These predictors included nighttime lights, vegetation indices, roads, surface texture, and topography. Satellite data used included products from the Suomi National Polar-Orbiting Partnership (Suomi NPP) spacecraft, Sentinal-2, and Sentinel-1. Preprocessing, analysis, and data aggregation was done with Google Earth Engine. Statistical analysis was used to analyze the correlation between predictors and OSM building footprints and a regression with Random Forests evaluated the potential of the remotely sensed variables to predict the presence of building footprints in a given area.

More detail about the study and the methodology used can be found in the published open access article:
Goldblatt, R., Jones, N., & Mannix, J. (2020). Assessing OpenStreetMap Completeness for Management of Natural Disaster by Means of Remote Sensing: A Case Study of Three Small Island States (Haiti, Dominica and St. Lucia). Remote Sensing, 12(1), 118. https://doi.org/10.3390/rs12010118
Related
- Using Crowdsourcing and Deep Learning to Map Human Settlements
- Transforming Satellite Imagery Classification with Deep Learning
Free weekly newsletter
Fill out your e-mail address to receive our newsletter!
By entering your email address you agree to receive our newsletter and agree with our privacy policy.
You may unsubscribe at any time.