Assessment of Urban Heat Island Patterns in Bangkok Metropolitan Area Using Time-Series of LANDSAT Thermal Infrared Data

Abstract

Bangkok is a rapidly expanding city with existing natural areas being replaced by developed areas creating an urban heat island (UHI) phenomenon in the city. LANDSAT imagery, near-infrared wavelength data, and time series information were used to study and to monitor the phenomenon of surface urban heat island (SUHI) in Bangkok. The variation of land surface temperature (LST) and the urban heat island intensity (UHII) phenomenon during 2008-2014 were investigated and the relationship between the UHII phenomenon and urban sprawl in Bangkok was studied. Using the UHII, we compared nine LST images of the investigated areas defined as inner, urban fringe and suburb zones. The UHI in Bangkok in the winter (dry) is higher than in the summer. Satellite imageries were used to classify the land use types as open spaces with high-rise buildings, very high density of buildings and high-rise buildings. Low vegetation index was found in urban fringe areas and inner city area with high surface temperature. The vegetation index value is high in areas of agricultural land, and low density building, it appears in suburb areas with low surface temperature. The results indicate that NDVI and High-rise building zones influence LST distribution and UHII phenomenon.

Landsat Time-series Images-based Urban Heat Island Analysis: The Effects of Changes in Vegetation and Built-up Land on Land Surface Temperature in Summer in the Hanoi Metropolitan Area, Vietnam

“…This indicates LSTs were negatively correlated with vegetation cover. However, opposite to those reported in previous studies (Liu and Zhang, 2011;Ranagalage et al, 2017;Sanecharoen et al, 2019), a significant positive regression coefficient of 0.56 was estimated in 2007 ( Figure 9 (b)). In addition, low values of coefficient of determination (R 2 ) of 0.076 in 1996, 0.001 in 2007 and 0.044 in 2016 indicate a relative bad goodness of fit for the observations.…”

“…These regression coefficients were negative and statistically highly significant (p<0.001) at the three-time points showing the LST is strongly and negatively correlated with vegetation cover. Our finding is consistent with those reported in previous studies (Kumar and Shekhar, 2015;Sanecharoen et al, 2019).…”

“…It can be concluded that there existed a very strong positive correlation and relationship between LST and builtup land during the period of 1996-2016. This finding is consistent with those reported in previous studies (Sanecharoen et al, 2019;Zhang et al, 2009), especially in the HMA .…”

“…The first studies of UHIs based on satellite-derived LST retrieval were carried out mainly using low-resolution NOAA AVHRR images (Balling and Brazel, 1988;Gallo et al, 1993). Since remotely sensed images of high resolution thermal infrared (TIR) satellite sensors such as 120-m resolution Landsat MSS and TM, 60-m resolution ETM+, and recently 100-m resolution TIRS started to be freely distributed, these data have also been widely utilized to derive LSTs for UHI studies (Liu and Zhang, 2011;Lu et al, 2009;Ranagalage et al, 2017;Ongsomwang et al, 2018;Sanecharoen et al, 2019). With the help of geographic information systems (GIS) techniques, many previous SUHI studies have successfully estimated the effects of urban landscape composition and pattern on LST (Ranagalage et al, 2017;Bokaie et al, 2016;Estoque et al, 2017;Kikon et al, 2016;Liu and Zhang, 2011;Weng et al, 2004).…”

“…Later, both these indices were used as land cover types (Zhang et al, 2009). Recent studies have shown that there exists a negative correlation of LST with NDVI and a positive correlation of LST with NDBI (Liu and Zhang, 2011;Ranagalage et al, 2017;Sanecharoen et al, 2019). In spite of these significant contributions, the main limitation of previous work is that the used method, the ULRM, fails to estimate the interrelationship between variables, thus, leading to low accuracy of the predicted model in the evaluation of the joint effect of both vegetation cover and built-up land on LST.…”