Integrating LandTrendr and historical maps for forest age classification, and spectral indices-based ecological assessment in urban and peri-urban Aichi, Japan

Effective planning and management of urban and peri-urban forests are increasingly critical in the context of rapid urbanization and climate change. Forest regeneration age strongly influences ecological conditions and is therefore crucial for effective management. However, the age dynamics of urban forests, which are often fragmented, species diverse, and poorly documented, remain underexplored. In this study, we examined age related ecological dynamics across a 2,198 km² forested landscape in Aichi Prefecture, Japan, including Nagoya City and surrounding municipalities. The forest regeneration history from 1986 to 2022 for current existing forests (extracted from the 2022 High Resolution Land Use Land Cover map) was reconstructed using the LandTrendr algorithm based on the Landsat dataset and complemented by historical forest maps (1955 and 1977). Forests were classified into four regeneration age groups: Group 1 (post-1986), Group 2 (1977–1986), Group 3(1955–1977), and Group 4(pre-1955). Classification accuracy was validated through field surveys and aerial imagery. Clear evidence for post-1986 and > 85% accuracy for older stands was obtained. Polygonization of age groups enabled spatial fragmentation assessment. It confirmed the effectiveness of the algorithm even in small heterogeneous patches, highlighting the advantage of integrating multi-source datasets for urban contexts. An ecological assessment using four spectral vegetation indices, i.e., normalized burn ratio (NBR), normalized difference vegetation index (NDVI), normalized difference water index (NDWI), and tasseled cap brightness (TCB), was conducted after independent validation and pre-selection. The results revealed apparent age-related differences: younger stands were highly fragmented and exhibited greater variability in spectral indices, with Group 1 consistency lower in NBR and NDVI, but higher in NDWI and TCB, and the value also varied, reflecting sparse canopies, reduced water retention, and heterogeneous phenology, whereas older forests formed larger, contiguous patches with stable canopy structures and consistent spectral signatures. Seasonal analysis indicated that autumn provided a robust differentiation between younger and older stands. Further integration of forest type data (evergreen broadleaf forest (EBF), deciduous broadleaf forest (DBF), and evergreen needleleaf forest (ENF)) showed that forest type amplified age-related spectral differences. These findings demonstrate the potential of combining time-series algorithms with multi-source datasets to reconstruct urban forest histories, and underscore the ecological importance of forest regeneration age and type in shaping spatial and seasonal forest dynamics.