Modelling the Impact of Land Cover Change on Urban Heat/Cool Island of Bengaluru Metropolitan City

Abstract

Urbanization has emerged as the most drastic and irreversible form of human-induced landscape change. Rise in temperature in urban area leads to high building energy consumption and degraded environmental qualities in the built environment. Hence, Urban Heat Island (UHI) effect has emerged as a key research top in the field of urban ecology and urban climatology. In most of the developing countries, man-made developments in the environment have led to the growing demand to contextualize the Land Use Land Cover (LULC) changes and Land Surface Temperature (LST) variations. Due to the modification in the surface properties of the cities, a difference in energy balance between the cities and its non-urban surroundings is observed. The present study was focussed on the analysis of spatial and temporal patterns of LULC and LST and its interrelationship in Bengaluru Urban district, India during the period from 1989 to 2017 using remote sensing data. Bengaluru is one of the rapidly growing cities in India and there is an urgent need for investigating the spatio-temporal patterns of LULC and LST in the region. The datasets used for the study mainly comprises of Landsat images and MODIS data from 1989 to 2020. The land cover maps of the study area were prepared for the years 1989, 1994, 2001, 2005, 2014 and 2017 using supervised classification. Intensity analysis was performed for the interval to analyse the LULC change and identify the driving forces. The impact of land cover change on LST was assessed using hot spot analysis (Getis-Ord Gi* statistics). The results of this study show that (a) dominant land cover change experienced is the increase in urban area (approximately 40%) and the rate of land cover change was faster in the time period 1989-2001 than 2001-2017. (b) the major transition witnessed is from barren and agricultural land to urban (c) Over the period of 28 years, LST patterns for different land cover classes exhibit an increasing trend with an overall increase of approximately 6ºC and the mean LST of urban area increased by about 8ºC (d) LST pattern change can be effectively analysed using hot spot analysis (e) As the urban expansion occurs, the cold spots have increased, and it is mainly clustered in theurban area. It confirms the presence of an urban cool island effect in Bengaluru urban district. LST and land cover interaction was modelled in a comprehensive and efficient way in the semi-arid tropical metropolitan city. Even though this interaction has been discussed widely in many literatures, the study facilitates the modelling and parameterization of LST and urban growth in an adequate way. Spatial distribution of LST and land cover types of the area were examined in the circumferential direction, and the contribution of land cover classes on LST was studied over 28 years. Urban growth and LST were modelled using Landsat and MODIS (Moderate Resolution Imaging Spectroradiometer) data for the years 1989, 2001, 2005 and 2017 based on the concentric ring approach. The study provides an efficient methodology for modelling and parameterization of LST and urban growth by fitting an inverse S-curve into Urban Density (UD) and mean LST data. In addition, Multiple Linear Regression (MLR) models which could effectively predict the LST distribution based on surface area ratios were developed for both day and night time. Further, the relationship between land cover types such as urban, vegetation, water and LST is determined for different years emphasizing the impact of land cover change on the daytime and night time surface heating. The non-linear relationship between surface area ratios and LST was established using a hybrid Particle Swarm Optimization - Support Vector Regression (PSO-SVR) model for the years 1989, 2001, 2005 and 2017. Based on the analysis of remotely sensed data for LST, it is observed that over the years, urban core area has increased circumferentially from 5 km to 10 km, and the urban growth has spread towards outskirts beyond 15 km from the city centre. As urban expansion occurs, the area under the study experiences an expansive cooling effect during day time; at night, an expansive heating effect is experienced in accordance with the growth in UD in the suburban area and outskirts. The regression models that were developed have relatively high accuracy with R2 value of more than 0.94 and could explain the relationship between LST and land cover types. The study also revealed that there exists a negative correlation between urban, vegetation, water body and LST during day time while a positive correlation is observed during night.The values of the statistical indices prove the feasibility and efficacy of PSO algorithm in tuning the hyperparameters of SVR. The Hybrid PSO-SVR model was built on the tuned hyperparameters for modelling LST with different surface area ratios at different time frames. For surface area ratio, R2 value in the range of 0.94 and 0.97 was obtained for MLR and Hybrid PSO-SVR model respectively. The spatio-temporal variation of urban surface characteristics and its relationship with LST was also modelled over the period from 1989 to 2017. Remote sensing indices such as NDVI (Normalized Difference Vegetation Index), NDWI (Normalized Difference Water Index), and NDBI (Normalized Difference Built-up Index), was determined from Landsat images for the years 1989, 2001, 2005 and 2017. Linear relationship between LST and these remote sensing indices were studied by employing MLR technique. Further, the proposed Hybrid PSO-SVR model was applied to the datasets to predict the values of LST based on these remote sensing indices. Hypothetical scenarios were introduced in the prediction to assess the impact of change in vegetation and water bodies on LST. Temporal variation of urban heat anomaly of the region over the period of study was also investigated. NDBI has drastically increased in the year 2017 which is caused by the increase in barren land and urban areas while NDVI and NDWI has decreased over the years. Higher values of NDBI are scattered in the outskirts while higher NDVI and NDWI values are distributed in the urban centre. R2 value in the range of 0.80 and 0.85 was obtained for MLR and Hybrid PSO-SVR model respectively. Hybrid PSO-SVR model proved to be effective in establishing the relationship between LST and urban surface characteristics, NDVI, NDBI and NDWI and in predicting the future LST. From the hypothetical scenario analysis, it can be concluded that introduction of vegetation and water bodies in the suburban and urban fringes will reduce the difference in LST between urban and rural areas. The magnitude of urban heat anomaly can be curtailed by developing green corridors and artificial lakes in the suburban and urban fringes of Bengaluru.Thus, this study could assist urban planners and policymakers in understanding the scientific basis of urban heating effect and predict LST for the future implementation of green infrastructure. The findings of this work can be used as a scientific basis for the sustainable development and land use planning of the region in the future. The proposed methodology could be applied to other urban areas for quantifying the distribution of LST and different land cover types and their interrelationships.