The wildland–urban interface (WUI) is where buildings and wildland vegetation meet or intermingle1,2. It is where human–environmental conflicts and risks can be concentrated, including the loss of houses and lives to wildfire, habitat loss and fragmentation and the spread of zoonotic diseases3. However, a global analysis of the WUI has been lacking. Here, we present a global map of the 2020 WUI at 10 m resolution using a globally consistent and validated approach based on remote sensing-derived datasets of building area4 and wildland vegetation5. We show that the WUI is a global phenomenon, identify many previously undocumented WUI hotspots and highlight the wide range of population density, land cover types and biomass levels in different parts of the global WUI. The WUI covers only 4.7% of the land surface but is home to nearly half its population (3.5 billion). The WUI is especially widespread in Europe (15% of the land area) and the temperate broadleaf and mixed forests biome (18%). Of all people living near 2003–2020 wildfires (0.4 billion), two thirds have their home in the WUI, most of them in Africa (150 million). Given that wildfire activity is predicted to increase because of climate change in many regions6, there is a need to understand housing growth and vegetation patterns as drivers of WUI change.

Background Wildland–urban interface (WUI) maps identify areas with wildfire risk, but they are often outdated owing to the lack of building data. Convolutional neural networks (CNNs) can extract building locations from remote sensing data, but their accuracy in WUI areas is unknown. Additionally, CNNs are computationally intensive and technically complex, making them challenging for end-users, such as those who use or create WUI maps, to apply. Aims We identified buildings pre- and post-wildfire and estimated building destruction for three California wildfires: Camp, Tubbs and Woolsey. Methods We evaluated a CNN-based building dataset and a CNN model from a separate commercial vendor to detect buildings from high-resolution imagery. This dataset and model represent to end-users the state of the art of what is readily available for potential WUI mapping. Key results We found moderate accuracies for the building dataset and the CNN model and a severe underestimation of buildings and their destruction rates where trees occluded buildings. The CNN model performed best post-fire with accuracies ≥73%. Conclusions Existing CNNs may be used with moderate accuracy for identifying individual buildings post-fire and mapping the extent of the WUI. The implications are, however, that CNNs are too inaccurate for post-fire damage assessments or building counts in the WUI.