Relationship between urban landscape structure and land surface temperature: Spatial hierarchy and interaction effects

Urban landscape structure has profound impacts on the urban thermal environment (UTE), however, the spatial hierarchy and interaction effects of the relationship between them are limitedly explored. Taking Xi'an Metropolitan area as the case, this study applies satellite imageries, local contour tree algorithm, and XGBoost model to identify the spatial hierarchy of UTE represented by nested heat centers, and then elaborate the relationship between landscape structure (landscape metrics) and land surface temperature (LST) across the spatial levels of UTE. We identify 21 main urban heat island (UHI) areas and their UTE's hierarchy, among which the Xi'an main urban district has the most complex structure (six levels) and is the core area. Then, in the core area, we found that the discussed relationship varies across the levels of UTE. Specifically, the importance of impervious surface increases from the local heat center with the lowest level (40.7%) to the entire UHI area with the highest level (62.0%), whereas that of urban green space has opposite trend (from 41.4% to 25.6%). Besides, landscape metrics within specific range of values show significant interaction effects on LST. Our results imply that differential landscape optimisation criteria should be considered in the different spatial levels of UTE.