Investigating the association between the spatial patterns of vegetation coverage and urban thermal environment

The impact of the increasing temperature on the built environment has been widely researched and documented. One of the critical intervention strategies  is to increase the amount of vegetation in cities. However, how various vegetation types and their composition and configuration could effectively cool temperature hasn't been fully investigated. This thesis aims to fill this gap by addressing the following objectives: (i) to analyse the association between Land Surface Temperature (LST) and air temperature and to investigate the impact of environmental and climatic factors on their association; (ii) to identify and measure essential landscape metrics that can describe the composition and configuration of vegetation types (tree, grass, shrubs); and (iii) to analyse geostatistical association between the spatial patterns of vegetation types, LST and air temperature. To achieve objective one, this research first developed an automated cloud-computing algorithm in the Google Earth Engine (GEE) platform that generates diurnal and nocturnal (LST) at spatial resolution of 30m  from Landsat-8 and MODIS satellite imageries using      the Random-Forest (RF) machine learning technique. The WunderGround crowdsourced database and available weather stations from the Bureau of Meteorology (BOM) were used to derive air temperature data 30m spatial resolution. The air temperature datasets were generated by applying the Empirical Bayesian Kriging (EBK) interpolation method .      The Urban Monitoring data™ of greater Melbourne with the spatial resolution of 20 cm was used as the primary vegetation cover data to address objective two of this research. The vegetation heterogeneity and functionality concerning the LST and air temperature, investigated by analysing their spatial composition and configuration. Measuring the quantity of the spatial patterns were conducted by landscape metric algorithms. Landscape metrics are measurable units of landscape spatial patterns that describe ecological processes over time and space. After reviewing and quantifying the most effective interferer landscape metrics on LST and air temperature characteristics, the association between LST, air temperature and spatial patterns of the vegetation types were investigated by utilising raster-based Pearson correlation analysis method. This research indicated  that the LST and air temperature were positively associated. The coefficient value (r) and coefficient of determination (R2) values indicated that the strength of their association is highest during the early afternoon (r: 0.77-R2: 0.602) and lowest at night-time (r:0.05-R2: 0.002). Numerous climatic and environmental factors influence the spatial distribution of the LST and air temperature in Greater Melbourne during early afternoon and night-time. This research showing that during the early afternoon, the following factors resulted in LST, and air temperature variation while having a direct association with them: (i) Soil Moisture Content (SMC), (ii) NDBI and (iii) windspeed. In contrast, some other factors influence of LST and air temperature which resulted in their reduction. These factors are humidity and NDWI. During the night-time, SMC ,NDBI and NDWI influence on the LST and have a direct association with that. However, humidity, NDVI and windspeed have resulted in LST reduction. The situation for the night-time air temperature is different as NDBI, SMC, and NDVI has a direct association with air temperature increment. In contrast, humidity, windspeed, and NDWI has a inverse association with air temperature during the night. The findings from objective two showed that the Patch Density (PD), Edge Density (ED), Shannon's diversity index (SHDI), Patch Compactness (PC) and Fractal Dimension Index (FRAC) are the prevalent configuration and composition landscape metrics related to the urban vegetation types that can explain the LST and air temperature. The individual analysis between each landscape metric, LST and air temperature indicated that shrubs' ED was an important configuration metric that has a substantial to very strong inverse association with LST during early afternoon (r: - 0.63- R2:0.39). Next important configuration metric that has a substantial to very strong inverse association with air temperature and LST during early afternoon is tree PD (r: -0.62 R2:0.38). The calculated coefficient-values are mainly negative in most areas of Greater Melbourne, except north, northwest and a few areas in the inner-southern-city (r >0). Other important landscape metric was tree ED which had a moderate to substantial inverse association with both air temperature (r: - 0.49- R2:0.24) and LST (r: - 0.42- R2:0.17).  The spatial distribution of the coefficient-value between grass PD and LST and air temperature showed that most areas had a negative coefficient-value (r >0), except inner-metro and inner-east areas. Furthermore, the correlation coefficient value between grass PD and early afternoon LST is strongest association (r: -0.19- R2:0.03) in comparison to other time of the day and night regarding LST and air temperature association with grass PD. The calculated r and R2 values between LST and tree height showed their moderate positive association during the night (r: 0.41-R2:0.16). Focusing on the composition metrics, for SHDI, the association between that, LST and air temperature are moderately with (r: -0.40 and R2:0.16) for LST and (r: - 0.34 R2:0.11) for air temperature. The coefficient-value map between LST and SHDI demonstrated that the coefficient range was mainly negative in the inner-metro, inner-east and northeast areas of Greater Melbourne, except a few areas in the inner west. This research highlights the importance of developing localised and city-wide urban vegetation strategies considering the geospatial association between LST and air temperature. Furthermore, the findings of this research can help planners and designers to implement urban greenery strategies while considering different geospatial, environmental, and climatic characteristics and places' parameters. One of the strategies generated from the SHDI, LST and air temperature association could combine all types of urban vegetation (shrubs, trees, and grass) in areas of the city that the LST mostly has higher values during the early afternoon. Planting shrubs along with tall trees in dense-urban areas located in the inner-metro suburbs will reduce the temperature more effectively considering the effect of tree height and shrubs ED on LST during the early afternoon. Also, based on shrubs ED results and its role in reducing the air temperature and early afternoon LST, planting complex shapes of vegetation (shrubs mainly) was more beneficial in cooling the environment