Restricted Access
Reason: Access restricted by the author. A copy can be requested for private research and study by contacting your institution's library service. This copy cannot be republished
Informed implementation of greening as a heat mitigation measure in Melbourne, Australia: a remote sensing study
thesis
posted on 2017-02-21, 04:45 authored by Sultana Nasrin NuryUsing remote
sensing data, this research investigates the spatial relationship between land
surface characteristics and urban climate to assess the potential of increasing
vegetation cover to mitigation the urban heat island (UHI) in Melbourne,
Australia. This investigation included the estimation of the spatial
variability of land cover, morphometric parameters, land surface temperature
(LST) and the relevant urban fluxes of substrate (G), sensible (H) and
evaporative (λET) heat, along with relevant statistical analysis. This allowed
for a quantification of the influence of surface characteristics on the
distributed LST and surface energy fluxes. Three local government areas (LGA)
in the Melbourne metropolitan area, namely Melbourne, Monash and Darebin, were
selected for this research.
Within a short span of distance, urban areas are highly heterogeneous both horizontally and vertically. These heterogeneous surfaces play a significant role in energy partitioning in the urban area and as such, detailed land cover and morphometric information are essential for this research. For the first time in Melbourne, highly detailed land cover information and building morphometric parameters have been derived using LiDAR data. The land cover types include buildings, roads, other impervious surfaces, trees, grass and water. The accuracy of the derived data was assessed by comparing it with manually digitized reference land cover data from aerial photographs. The overall accuracy of the derived land cover is very high (94%) and closely replicates the in situ conditions.
Mean building and tree height were extracted for each building and tree polygon from LiDAR data and used as the basis for the estimation of morphometric parameters including plan area fraction (λp), density ratio (λB), frontal area index (λf), complete aspect ratio (λc), roughness length (Z0) and displacement height (Zd). These parameters have also been estimated for the first time for the study area and derived values are in accordance with the few published values for other cities.
The spatially distributed surface temperature and heat fluxes for the study area were estimated using Landsat TM5 at the time of the satellite overpass. To understand the afternoon and night time LST distribution, eight-day composite MODIS aqua LST images for the month of January from 2005 to 2010 were used for the study area. The areas having higher LST than the estimated mean represent relative urban warmth and considered the most probable areas prone to excessive heat, also termed surface urban heat islands (SUHI). In the Melbourne LGA, nearly 44% of the area showed positive SUHI intensity from Landsat TM5 LST analysis, whereas in Monash and Darebin this positive intensity covered less than 8% and 15% of the area respectively. The central part of Melbourne displayed a night time SUHI, whereas in Darebin and Monash, this effect was negligible.
Surface cover and morphometry heavily influenced the spatial distribution of Landsat derive LST and energy fluxes. In all cases, with increasing imperviousness LST increased along with the SUHI effect, whereas with increasing vegetation cover LST decreased. In most cases, LST increased with increasing λp, λB, λf, and λ, except for high-rise building areas where building shadows appears to reduce the LST.
The spatial distribution of energy fluxes was estimated using an energy balance model utilising two Landsat TM5 images for the Melbourne metropolitan area, derived morphometric parameters and climate data. The estimated energy fluxes showed the significant effect of land cover variability. Impervious cover was the dominating factor for the spatial distribution of storage (G) and sensible (H) heat fluxes. In all three LGAs, the value of G increased with increasing impervious cover and decreased with increasing vegetation cover, but the effect is more evident for tree than for grass cover. In all-three study areas, the spatial distribution of H was highly variable.
The evapotranspiration (ET) rate also varies significantly for grass and trees. Grass cover always showed higher mean LST even when it has higher mean ET values. The LST and ET were inversely correlated for both grass and trees and the coefficient of determination (R2) is more than 0.9 for trees in all three areas, whereas for grass it is more variable and ranges between 0.2 to 0.9. From statistical analysis it is evident that increases in ET rate and associated decreases in LST are higher for trees than grass, indicating that trees are more efficient than grass for LST reduction in the urban environment of Melbourne. As LST has a direct influence on air temperature, trees will be most effective to reduce air temperature and mitigate excess urban heat.
This study demonstrated that remote sensing can be used as a tool to inform the implementation of vegetation as a heat mitigation measure in an urban area. Remote sensing can be used to characterize the land surface, derive estimates of LST and provide input data for energy balance modelling. These estimated and modelled parameters and their spatial relationships can be used to evaluate future land use scenarios, and can also be used in prognostic climate modelling. The resulting relationships can also facilitate the planning of a balanced spatial arrangement of land use types where land uses that are prone to excessive heat can be managed in association with relatively cooler land uses.
Globally, many researchers are contributing to an improved understanding of how LST and land cover/land-use types are related using a variety of data sources and methods. Many of the outcomes of this research are location specific, so further research is recommended to apply these methodologies and correlation values in other applications and in other places. The outcome of this thesis is based on the data from three LGAs of the Melbourne metropolitan area, and will add to the global understanding. It is recommended that similar studies are conducted for the other local government areas in Melbourne as well.
Within a short span of distance, urban areas are highly heterogeneous both horizontally and vertically. These heterogeneous surfaces play a significant role in energy partitioning in the urban area and as such, detailed land cover and morphometric information are essential for this research. For the first time in Melbourne, highly detailed land cover information and building morphometric parameters have been derived using LiDAR data. The land cover types include buildings, roads, other impervious surfaces, trees, grass and water. The accuracy of the derived data was assessed by comparing it with manually digitized reference land cover data from aerial photographs. The overall accuracy of the derived land cover is very high (94%) and closely replicates the in situ conditions.
Mean building and tree height were extracted for each building and tree polygon from LiDAR data and used as the basis for the estimation of morphometric parameters including plan area fraction (λp), density ratio (λB), frontal area index (λf), complete aspect ratio (λc), roughness length (Z0) and displacement height (Zd). These parameters have also been estimated for the first time for the study area and derived values are in accordance with the few published values for other cities.
The spatially distributed surface temperature and heat fluxes for the study area were estimated using Landsat TM5 at the time of the satellite overpass. To understand the afternoon and night time LST distribution, eight-day composite MODIS aqua LST images for the month of January from 2005 to 2010 were used for the study area. The areas having higher LST than the estimated mean represent relative urban warmth and considered the most probable areas prone to excessive heat, also termed surface urban heat islands (SUHI). In the Melbourne LGA, nearly 44% of the area showed positive SUHI intensity from Landsat TM5 LST analysis, whereas in Monash and Darebin this positive intensity covered less than 8% and 15% of the area respectively. The central part of Melbourne displayed a night time SUHI, whereas in Darebin and Monash, this effect was negligible.
Surface cover and morphometry heavily influenced the spatial distribution of Landsat derive LST and energy fluxes. In all cases, with increasing imperviousness LST increased along with the SUHI effect, whereas with increasing vegetation cover LST decreased. In most cases, LST increased with increasing λp, λB, λf, and λ, except for high-rise building areas where building shadows appears to reduce the LST.
The spatial distribution of energy fluxes was estimated using an energy balance model utilising two Landsat TM5 images for the Melbourne metropolitan area, derived morphometric parameters and climate data. The estimated energy fluxes showed the significant effect of land cover variability. Impervious cover was the dominating factor for the spatial distribution of storage (G) and sensible (H) heat fluxes. In all three LGAs, the value of G increased with increasing impervious cover and decreased with increasing vegetation cover, but the effect is more evident for tree than for grass cover. In all-three study areas, the spatial distribution of H was highly variable.
The evapotranspiration (ET) rate also varies significantly for grass and trees. Grass cover always showed higher mean LST even when it has higher mean ET values. The LST and ET were inversely correlated for both grass and trees and the coefficient of determination (R2) is more than 0.9 for trees in all three areas, whereas for grass it is more variable and ranges between 0.2 to 0.9. From statistical analysis it is evident that increases in ET rate and associated decreases in LST are higher for trees than grass, indicating that trees are more efficient than grass for LST reduction in the urban environment of Melbourne. As LST has a direct influence on air temperature, trees will be most effective to reduce air temperature and mitigate excess urban heat.
This study demonstrated that remote sensing can be used as a tool to inform the implementation of vegetation as a heat mitigation measure in an urban area. Remote sensing can be used to characterize the land surface, derive estimates of LST and provide input data for energy balance modelling. These estimated and modelled parameters and their spatial relationships can be used to evaluate future land use scenarios, and can also be used in prognostic climate modelling. The resulting relationships can also facilitate the planning of a balanced spatial arrangement of land use types where land uses that are prone to excessive heat can be managed in association with relatively cooler land uses.
Globally, many researchers are contributing to an improved understanding of how LST and land cover/land-use types are related using a variety of data sources and methods. Many of the outcomes of this research are location specific, so further research is recommended to apply these methodologies and correlation values in other applications and in other places. The outcome of this thesis is based on the data from three LGAs of the Melbourne metropolitan area, and will add to the global understanding. It is recommended that similar studies are conducted for the other local government areas in Melbourne as well.
