Abstract
Urban green spaces are essential for regulating land surface temperature (LST), but current research frequently neglects their structural complexity and perceived accessibility by humans. To bridge this gap, our study utilizes two complimentary metrics: the satellite-derived Normalized Difference Vegetation Index (NDVI) and the street-level Green View Index (GVI), both employed to assess Guangzhou's urban thermal environment. Distinct statistical and spatial distribution patterns of NDVI and GVI were identified among districts in Guangzhou, China. NDVI values varied between 0.12 and 0.64, whereas GVI values ranged from 0.18 to 0.47. The LST varied from 27.61 to 41.99 °C, with a global Moran's I of 0.96 signifying robust spatial autocorrelation. To evaluate the impact of urban morphology on LST, we employed three regression models, with the multiscale geographically weighted regression (MGWR) demonstrating superior performance, with R(2) = 0.727, AICc = 2185.43, and RSS = 328.11. Regression results revealed that building density (BD) and average building volume (BV) are positively connected with LST. In contrast, GVI and NDVI exhibit negative associations. This study integrates vertical (NDVI) and horizontal (GVI) greenery viewpoints with urban morphological characteristics, offering actionable insights for urban planners to enhance green infrastructure and more effectively offset the urban heat island effect.