The relationships between 2D and 3D green index altered by spatial attributes at high spatial resolution

• Published in: Urban forestry & urban greening Vol. 101; p. 128540
• Main Authors: Bai, Zhuhui, Wang, Zhifang, Li, Dan, Wang, Xiaomeng, Jian, Yuqing
• Format: Journal Article
• Language: English
• Published: Elsevier GmbH 01.11.2024
• Subjects: Green index; Green View Index; Moderation model; Tree cover density; Vegetation composition; Green View Index; Green index; Vegetation composition; Moderation model; Tree cover density
• ISSN: 1618-8667
• DOI: 10.1016/j.ufug.2024.128540

Understanding the relationship between 2D and 3D green indices holds significant implications for governance, especially amid the escalating demand for urban green spaces during rapid urbanization. However, the existing studies reveal inconsistent results pertaining to the relationships between 2D and 3D green indices and lack a systematic examination of spatial attributes' influence on the relationships. This study addresses these gaps by 1) assessing the correlation between 2D and 3D green indices using high-resolution data, 2) identifying spatial attributes that alter the 2D-3D association, and 3) determining the thresholds of spatial attributes that significantly impact the 2D-3D relationships. The Moderation Model was initially introduced to evaluate the mechanisms within its respective field. High-resolution data was collected from a campus, focusing on key spatial attributes: tree cover density (TC), plot ratio (PR), and vegetation composition (VC). NDVI and GVI were chosen to represent the 2D and 3D green index. The results indicate a significant correlation between NDVI and GVI (R=0.478, p<0.005). TC was identified as the fundamental moderator of the association. A medium TC (32.46 %) was the threshold for a significant NDVI - GVI correlation. Only at low TC (0–10.82 %), interactive effects between spatial attributes were observed: VC (type (e)) substituted NDVI to affect GVI directly; Above a PR of 1.95, the correlation between NDVI and GVI diminishes. In general, NDVI cannot substitutes GVI in areas where TC is higher than 32.46 %, or in areas where TC is lower than 10.82 % with many tall buildings (higher than 8 floors). The findings innovatively reveal the dynamic variation and thresholds in the 2D-3D green index relationships based on site-specific attributes rather than the resolution of data. Practically, a decision-making tree of green index governance is then proposed for relevant stakeholders to choose according to different site conditions.