Modeling the intensity of surface urban heat island based on the impervious surface area

Abstract

The urban heat island (UHI) effect is anticipated to be intensified and cause severe damage to human society under future urbanization and climate change. Due to the complicated human-nature interactions in the urban domain, it is challenging to characterize the intensity of UHI quantitatively, especially at the global scale. Here, we proposed a framework to estimate the surface urban heat island intensity (SUHII) based on satellite-derived land surface temperature (LST) and the derived impervious surface area (ISA) data across 355 cities worldwide from 2003 to 2018. First, we employed linear regression models to estimate LST using ISA as the explanatory variable, given that the LST generally elevated with increasing urban intensity along the urban-rural gradient. Then, using the modeled LST, we estimated the spatiotemporal patterns of SUHII worldwide. The established linear regression models with ISA as the explanatory variable perform reasonably well in estimating LST for both daytime and nighttime, with average RMSEs of 1.40 and 0.80 °C, respectively. The resulting SUHII reasonably captures the spatial heterogeneity across cities in the world, with the highest value in the tropical region at daytime and in the arid region at nighttime. The estimated and observed SUHII show good agreement with average R2 values of 0.81 and 0.72 for daytime and nighttime, respectively. Using ISA as an indicator for LST estimation, this study first implemented a comprehensive estimation of SUHII at the global scale. The proposed methodology is of great potential to project future SUHII by expanding urban impervious surfaces under diverse scenarios jointly determined by shared socioeconomic pathways and climate change scenarios.