Time-Series Analysis Reveals Intensified Urban Heat Island Effects but without Significant Urban Warming

Jia Wang,Jing Wang,Weiqi Zhou

doi:10.3390/rs11192229

Abstract

Numerous studies have shown an increased surface urban heat island intensity (SUHII) in many cities with urban expansion. Few studies, however, have investigated whether such intensification is mainly caused by urban warming, the cooling of surrounding nonurban regions, or the different rates of warming/cooling between urban and nonurban areas. This study aims to fill that gap using Beijing, China, as a case study. We first examined the temporal trends of SUHII in Beijing and then compared the magnitude of the land surface temperature (LST) trend in urban and nonurban areas. We further detected the temporal trend of LST (TrendLST) at the pixel level and explored its linkage to the temporal trends of EVI (TrendEVI) and NDBI (TrendNDBI). We used MODIS data from 2000 to 2015. We found that (1) SUHII significantly increased from 4.35 °C to 6.02 °C, showing an intensified surface urban heat island (SUHI) effect, with an annual increase rate of 0.13 °C in summer during the daytime and 0.04 °C in summer at night. In addition, the intensification of SUHII was more prominent in new urban areas (NUA). (2) The intensified SUHII, however, was largely caused by substantial cooling effects in nonurban areas (NoUA), not substantial warming in urban areas. (3) Spatially, there were large spatial variations in significant warming and cooling spots over the entire study area, which were related to TrendNDBI and TrendEVI. TrendNDBI significantly affected TrendLST in a positive way, while the TrendEVI had a significant positive effect (p = 0.023) on TrendLST only when EVI had an increasing trend. Our study underscores the importance of quantifying and comparing the changes in LST in both urban and nonurban areas when investigating changes in SUHII using time-series trend analysis. Such analysis can provide insights into promoting city-based urban heat mitigation strategies which focused on both urban and nonurban areas.

Highlights

More than 50% of the world population is living in cities [1]
After we obtained the data of surface urban heat island intensity (SUHII), land surface temperature (LST), enhanced vegetation index (EVI), and normalized difference build-up index (NDBI) from 2000 to 2015, we investigated the temporal trends of such variables (e.g., SUHII) at every pixel using ordinary least squares (OLS) linear regression models [36,43,72]
Similar patterns of nighttime SUHII were observed for OUA and new urban areas (NUA), but OUA had a slightly weaker trend (Figure 4)

Summary

Introduction

More than 50% of the world population is living in cities [1]. Rapid urbanization results in remarked changes of natural land to developed land, leading to a series of environmental issues [2,3,4,5,6,7,8]. Urban heat island intensity (UHII), the difference in temperature (air temperature and/or land surface temperature, LST) between urban and rural areas [10,11], is one of the most important indicators that has been used to quantify the magnitude of the UHI effect [15]. (2) How does UHII and its spatiotemporal variation relate to the social and biophysical properties of urban and rural areas, such as urban size and urban–rural enhanced vegetation index (EVI) difference? UHII had a strong negative relationship with the difference between the urban and rural vegetation indexes [10,25,27,29,30,31], but a significantly positive relationship with urban–rural nighttime lights [26]

Objectives

Methods

Results

Discussion

Conclusion