Historic gardens are green spaces characterised by tree stands with several veteran specimens of high artistic and cultural value. Such valuable plant components have to cope with biotic and abiotic stress factors as well as ongoing senescence processes. Maintaining tree health is therefore crucial to preserve their ecosystem services, but also to protect the monument and visitor health. In this context, finding smart, fast and cost-effective management solutions to monitor health and detect critical conditions for both stands and individual veteran trees can promote garden conservation. For this reason, we developed a novel framework based on Sentinel2 imagery, LiDAR sources and automatic cameras to identify risk spots regarding trees in historic gardens. The pilot study area consists of two closed Italian gardens from the 16th century, which were analysed as a unique Historic Garden System (HGS). The tree health status at stand level was assessed using a criterion based on the Normalized Difference Vegetation Index weighed on tree volume (NDVIt) and validated by a visual crown defoliation assessment. At the tree level, the health status of four veteran trees defined by the NDVIt was also evaluated using green chromatic coordinates (GCC) obtained from digital images acquired by cameras at daily intervals during one growing season. The 33% of the tree population was classified as being in poor health, i.e. "at risk". Veteran trees classified as "at risk" showed an anticipation of phenological phases and a lower GCC compared to reference trees. Despite variability determined by Sentinel medium resolution, the proposed framework showed good accuracy (0.74) for monitoring historical gardens. The semi-automatic risk point mapping system tested here proved to be effective in facilitating the management of historic gardens, which in turn could be applied in the wider context of urban greening.
Keywords: Crown defoliation; Heritage site management; Image analysis; Remote sensing; Tree monitoring; Urban forests.
Copyright © 2024 Elsevier Inc. All rights reserved.