Estimating the density of urban trees in 1890s Leeds and Edinburgh using object detection on historical maps

A groundbreaking study reveals that Edinburgh had a higher density of urban trees per hectare (40 ± 1) compared to Leeds (37 ± 1) during the late 19th century, based on an analysis of historical Ordnance Survey maps from the 1890s. This is the first quantitative data for urban tree counts in these cities, providing unprecedented insights into historical urban forests. Researchers developed a novel machine learning approach using object detection to automatically extract tree symbols from digitized maps with high accuracy (F1-score of 0.945). The method, applicable to other UK towns and cities, offers valuable lessons on symbol extraction from historical maps while addressing challenges such as training data generation and map source limitations. These findings contribute to understanding urban greening trends over time and inform future tree planting strategies.

Smith, E. S., Fleet, C., King, S., Mackaness, W., Walker, H., & Scott, C. E. (2025). Estimating the density of urban trees in 1890s Leeds and Edinburgh using object detection on historical maps. Computers, Environment and Urban Systems, 115, 102219. https://doi.org/10.1016/j.compenvurbsys.2024.102219