Distinct influences of urban villages on urban heat islands: A case study in the Pearl River Delta, China

Abstract

Widely scattered urban villages (UVs) and increasingly serious urban heat islands (UHIs) are common urban problems in highly urbanized regions, especially in the developing countries. However, the influences of UVs on UHIs remain little understood. In this study, different methodologies are performed to retrieve land surface temperature (LST) from thermal bands and the nearest object-oriented method with spectral, texture, shape metrics using ZY-3 high-resolution satellite imagery, and road network data are used to extract UVs and other land-use types in the Guangzhou–Foshan (GF) core areas of Pearl River Delta (PRD). Moreover, the relationship between LST and land-use types is then analyzed on the multiple scales. The results show that five land-use types (vegetation, normal construction land (NCL), UVs, water, and unused land) extracted by the object-oriented method were qualified for subsequent analysis because of satisfactory overall accuracy (0.887) and the Kappa coefficient (0.863). In the GF core areas presenting the most outstanding UHI effect across the PRD region, about 60.5% of the total area is covered by the impervious surfaces, including NCL (50.4%) and UVs (10.1%). The average LST of UVs was 1.89–2.97◦C lower than that of NCL. According to the average contribution index of thermal effect and the Pearson’s correlation coefficients, UVs present a relatively lower contribution to UHI and a weaker warming effect than NCL, but possess a higher contribution to UHI and a stronger warming effect than other land-use types, resulting in some slightly lower LST-valleys in the UVs adjacent to the NCL and distinct LST-peaks of UVs close to vegetation and water on the surface temperature profile lines. This work increases our understanding of the relationship between increasingly serious UHIs and widely distributed UVs, and would be valuable for local authorities to monitor and improve urban environment in metropolitan regions.