Urban environments increasingly rely on trees as key components of climate adaptation and liveability strategies, while also requiring careful management of their interaction with subsurface infrastructure. Ground Penetrating Radar (GPR) is widely used to investigate shallow subsurface conditions and detect tree root systems. However, translating spatially distributed geophysical observations into forms that can support consistent interpretation at the asset level remains a key challenge in practice.

Building on a previously developed framework, GPR data are segmented into repeatable spatial units within the root zone of individual trees, including common urban species such as lime and sycamore, and analysed to derive indicators of subsurface conditions. Each unit is then assigned a classification reflecting relative levels of intervention, resulting in a spatially distributed representation of management priorities. In this study, the focus is extended to the subsequent interpretation stage by explicitly defining and applying alternative strategies to aggregate zone-level classifications into tree-level representations. These include majority-based, worst-case, and proportional approaches, each providing a different synthesis of the same underlying geophysical evidence.

The results show that different aggregation strategies applied to identical zone-level classifications can lead to markedly different tree-level interpretations. This demonstrates that the transition from spatially resolved evidence to asset-level interpretation is not neutral, but depends on how the aggregation is performed. Formalising this step, the study enhances transparency and reproducibility in GPR-based assessment workflows and provides a structured basis for integrating subsurface information into practical evaluation workflows, supporting consistent interpretation and prioritisation in urban tree and infrastructure management.