Research On Landscape Pattern And Multi-scale Characteristics Of Urban Heat Island

The purpose of this thesis is to analyze the spatial characteristics and scale effects of urban heat island landscapes,and explore the correlation between urban heat island landscape and surface parameters.The representative mountainous city of Chongqing and the plain city of Chengdu areselected,and the Landsat 8 OLI_TIRS images areused as the data sources.First,based on the principle of landscape ecology,the spatial distribution characteristics and spatial structure of urban heat island landscape in Chongqing and Chengdu research areas arequantitatively analyzed.Second,the scale effect of the heat island landscape in the above study area is analyzed by using the particle size transformation and the amplitude variation,which is analyzed from the two aspects of the type level index and the landscape level index,and the variation trend of the scale effect of the heat island landscape between different types of cities is discussed.Finally,based on the optimal research granularity of the second part of the study,the relationship model between the urban heat island landscape pattern and the surface parameters of Chongqing are constructed from the administrative unit scale,and the relationship between them is analyzed by geographically weighted regression model.The research contents and main conclusions are as follows:1.The radiated transfer equation is used to invert the surface temperature data of the study area.Based on the improved “mean-standard deviation” method,the surface temperature data is divided into four heat island landscapes.Then,according to the landscape index,the spatial distribution characteristics and spatial structure of the urban heat island landscape are quantitatively analyzed.The conclusions are:(1)The distribution of heat island in Chongqing is concentrated and clustered,while that in Chengdu is dispersed and uniform.The surface temperature in Chongqing is mainly concentrated at 25 °C ~ 50 °C,while Chengdu is mainly concentrated at 28 °C ~ 43 °C.The separation degree of LST in Chongqing is greater than that in Chengdu,and the gradient effect is more obvious.(2)As the landscape matrix of the two research areas,the non-heat island landscape accounted for 56.05% and 54.61% respectively,which belonged to the absolute dominant landscape in the region.The density of landscape patches in Chengdu varies greatly,reaching a maximum of 0.5482.Patches are interwoven and fragmented,and the shape of the weak heat island is the most complex.The degree of fragmentation of the heat island is higher than that of the Chongqing area.2.The urban heat island landscape is characterized by landscape pattern index.The response curves of the heat island landscape with the scale change at the type level and landscape level are discussed from 18 kinds of grain size and 7 kinds of amplitudes,and the trend of the above-mentioned scale effect among different cities is further explored.The conclusions are:(1)At the class level,there are obvious granularity effects and magnitude effects in the four types of landscapes in NP,PD and LSI index.At the landscape level,the five landscape indices of PD,LSI,CONTAG,AI and SHDI all show obvious granularity effects,among which PD and The LSI index also show significant magnitude effects.(2)The optimal granularity of the granularity effect study is 30 meters.In the process of urban expansion from the central city,the "WHI-NHI-HI" heat island landscape distribution appears.In order to mitigate the impact of the heat island in the central city,more attention should be paid to the weak heat island rather than the traditional strong heat island.(3)The change trend of the index with obvious scale effect is close to the same among different research areas,and the spatial scale effect of the heat island landscape has obvious scale shift characteristics.3.Based on the geographically weighted regression model,the relationship between urban heat island landscape and surface parameters is discussed.It is concluded that the NDBI index has the highest interpretation rate for heat islands with obvious heat island effect in WHI,HI and SHI.Among the 18 groups of GWR models,11 models constructed by NDBI index have better fitting effects.NDBI has a high impact on the landscape superiority,landscape area and fragmentation of the heat island landscape.In addition,for areas where there is a significant urban heat island effect,NDBI has a significant positive correlation with the generation and enhancement of heat islands.At the landscape level,the NDBI index still has a high interpretation rate for the heat island landscape.Buildings in the city center have a greater impact on the degree of aggregation,spread and dominance of the heat island landscape,and there is a positive to a negative correlation trend from the center to the edge,while in the impact on landscape diversity,there is a negative to positive correlation trend.