Application of Spatial autocorrelation techniques in analyzing the heat island of Tehran

In this study, Landsat TM images were used to analyze the spatio-temporal changes of urban heat island of Tehran. The time period of images (13 images) is from 1986 to 2010. First, the mono-window algorithm was utilized to retrieve land surface temperature (LST) from thermal band of Landsat TM images. Then, the global and local spatial autocorrelations are used to quantitatively explore the spatial and temporal evolution of LST in Tehran. Results of global spatial autocorrelation showed that LST field in Tehran has changed to cluster pattern during the study period. In addition, the rate of global spatial autocorrelation increased from 1986 to 2010. According to the results of the global spatial autocorrelation two important remarks were obtained: (1) new thermal clusters were created during the study period. (2) The spatial extent of the old thermal clusters have been increased. To clarify the nature of the clusters (hot or cool cluster) and to explore the spatial variations of these clusters, the local spatial autocorrelation were used. These results confirmed the strengthening and extension of the hot clusters and weakening of the cool clusters. Exploration and Comparison of the spatial distribution of the high-high and low-low points during the study period showed that between 1986 and 2010 due to the degradation of half of the vegetation in the west and southwest of the Tehran (Districts of 18, 19, 21 and 22) cool clusters have been disappeared or fragmentized and new thermal clusters formed. As a result of these changes, urban heat island that was dominant in Mehrabad Airport (District 9) and portions of Districts 21 and 22 in 1986 expanded toward the west and southwest. Also, small thermal cluster have been emerged in the south and west of District 19. Moreover, because of the degradation of orchards in the northern part of Tehran (especially Shemiranat) cool clusters of these Districts have been weakened.