Historical Habitat Map of Switzerland (HIHAMAS)

Information about the spatial arrangement and extent of past habitats is crucial for understanding current biodiversity patterns and assessing the potential for ecological restoration. Over the past few decades, European landscapes have undergone significant changes due to increasing land-use intensity, with this trend accelerating post-World War II. In this study, we examine the feasibility of creating a 1946 habitat map for Switzerland that is compatible with contemporary habitat maps. This is achieved by choosing 8 case study areas that represent Switzerland's main biogeographical regions and charcteristic habitat types. Aerial photographs captured by the US Army Air Forces in the summer of 1946 have recently become available and from the base data for mapping. These high-resolution images (1-meter resolution) were segmented based on spectral and shape characteristics in the case study areas.

Manual interpretation of the orthoimages helped identify 15 distinct habitat classes, including wetlands, grasslands, arable lands, hedges, orchard meadows, and open forests. Using information from the black and white images as well as topography, climate, and anthropogenic influence, random forest models with active learning were trained to classify image segments.

The results showed that the highest class accuracies were achieved for habitats such as ‘Standing water,’ ‘Flowing water,’ ‘Glaciers, permanent ice and snow,’ and ‘Forests and other wooded land.’ However, lower user accuracies were noted for ‘Wetlands,’ ‘Hedges and tree rows,’ and ‘Buildings.’ The research demonstrates the effectiveness of object-based image analysis, machine learning, and diverse sampling approaches in deriving habitat maps from historical black-and-white aerial photography. These maps are instrumental in understanding the evolution of landscape ecology over time and can guide future conservation efforts.