Analysis of the Effect of Land Cover Spatial Configuration and Physical Characteristics of Buildings on the Surface Urban Cool Island Phenomenon

1. Introduction
The expansion of metropolitan areas has led to several physical, socioeconomic and environmental issues. One of the negative consequences is the intensification of urban heat islands (UHI) phenomenon in areas with more moderate climate. Studies on surface UHI have shown that in more arid environments there is a possibility for urban areas to have a lower temperature than the surroundings; a phenomenon that is called surface urban cool islands (SUCI) or surface urban heat sink.
Studies on the SUHI have shown that in addition to spatial composition and abundance of land covers, the shape of a land cover and its relative placement to other land covers (spatial configuration) is also important. These studies have pointed out that the land surface temperature (LST) decreases, as the complexity of urban patches in more moderate areas increase. However, studies on SUCI have mainly focused their attention on analyzing the effect of time of data acquisition and also the influence of the composition of land covers on the land surface temperature.
Another matter that has been somehow forgotten is the effect of building density and plot size on LST. Previous studies have noted the positive impact of the increase in building density on LST. However, the effect of plot size on the SUCI phenomenon has been significantly neglected.
Therefore, the purpose of this study is twofold. First, it tries to investigate the role of spatial configuration of land covers in the formation of SUCI in Shiraz urban area and then to assess the effect of urban density and plot size on the intensification on SUCI.
2. Materials and Methods
The process of the study involved three stages. First, the SUCI intensity for four different seasons was studied to determine if the SUCI phenomenon can be seen in Shiraz. Then, by conducting linear regression and with the use of landscape metrics, the relationship between the mean LST and the shape complexity and fragmentation of land covers was examined. Afterwards the effect of building density and plot size of the mean LST was studied.
Shiraz, the fifth largest city of Iran in terms of population with 1.869 million, has a hot semi-arid climate according to the Köppen climate classification.
2.1 Retrieval of LST
The methodology proposed by Sobrino et al. (2004) and Sobrino et al. (2008) were used to retrive LST for each year from the thermal bands of the images (band 6 of Landsat TM). It should be noted that the land surface emissivity was calculated for the thermal band with the simplified of Normalized Difference Vegetation Index (NDVI) Thresholds Method (NDVITHM).
2.2 SUCI intensity
A 10-km rural buffer around the core urban area which does not include water bodies were selected to compute the SUCI intensity.
2.3 Indicators to analyze the effect of landscape composition and spatial configuration and building characteristics on LST
The effect of landscape composition and spatial configuration on the intensification of SUCI was studied by using composition indices and landscape metrics. The composition indices, which are including NDVI and NDBI, measure the abundance of vegetation and non-vegetative land covers, respectively. Two metrics of Landscape Shape Index (LSI) and Fractal Dimension Index (FRAC) measure the shape complexity of land covers by studying the relative amount of perimeter per unit area, and Landscape Division Index (DIVISION) is the third spatial configuration index to study the degree to which land covers are broken up into separate patches.
The effect of building characteristics was also studied through examining the building density (Floor Area Ratio) and the size of plots in each census tract.
3. Discussion and Results
3.1 SUCI intensity
There were distinctive temperature differences between urban areas and the surroundings throughout the year. Also the SUCI phenomenon in Shiraz was at the highest level in summer. While the SUCI intensity was 6.62 degrees in summer, it was 3.42 degrees at the lowest level in winter.
3.2 The effect of landscape composition on LST
The value of NDVI showed that vegetation had no effect on the reduction of LST in urban area of Shiraz. Although a great deal of vegetation of the study area is within conservation areas in a few census tracts which had lower LSTs. On the other hand, the NDBI represented a significant effect of the built-up area on reducing land surface temperatures, contrary to studies on urban heat islands.
3.3 The effect of shape complexity and fragmentation of built-up areas on LST
Examining the landscape metrics showed that as the complexity and fragmentation of urban patches increased, the mean LST of census tracts rose as well. The increase in shape complexity and fragmentation of urban patches means these urban patches will have more adjacency with other land covers (such as vegetation or soil) on their edges.
3.4 The effect of building characteristic on the intensification of SUCI
The value of FAR pointed out that as the building density of each census tract increased, the mean LST decreased. On the other hand, the smaller the size of plots in a census tract, the lower the mean LST was. This can be clearly seen in historical areas of Shiraz in which the size of plots is smaller.
4. Conclusion
The results of this study revealed that the surface urban cool islands in Shiraz were significantly affected by the expansion of built-up areas. The analysis of landscape metrics showed as the built-up areas grew more irregularly, LST decreased. In other word, the shape complexity and fragmentation of urban patches led to an increase in LST. The shape complexity and fragmentation of urban patches means they have more adjacencies with other land covers such as vegetation. The analysis of building density and size of plots also showed the height of buildings and the size of plots affected the land surface temperature. In fact, the increase in building density and the decline in the size of plots led to the mitigation of LST and the intensification of SUCI. To sum up, while the expansion of built-up areas has a distinctive impact on mitigating LST, the SUCI phenomenon intensifies as this expansion occurs with higher density and in a smaller area.