Abstract: The phenomenon of urban heat island effect (UHIE) can synthetically indicate the environmental conditions of the city. Understanding the impact mechanism of urban landscape pattern  on the spatial distribution of thermal field  is significantly helpful to  optimize the layout of urban space and then improve environmental quality. Landsat-8 remote sensing image acquired on July 23, 2016 was used to derive land surface temperature (LST) and classify landscape types in Wuhan urbanized area.  Common landscape metrics were selected at patch level, patch type level and landscape level respectively to indicate landscape characteristics. Wuhan urbanized area was divided into 77 grids sized 3 km×3 km. In each grid, the related indexes and mean LST values were counted. Principal component regression equations between selected landscape metrics and LST were modeled to detect the influence mechanism of landscape pattern on UHIE. Results showed that: (1) About 76% of the selected indexes were significantly correlated with LST at 0.01 confidence level, and they also had strong correlation with each other. (2) The most efficient landscape indexes for LST variation were patch type area of Water body (CA_W), proportion of Construction Land in  Landscape area (PLAND_C), proportion of Core Area in Greenbelt of Landscape area (CPLAND_G), Mean Similarity Index of Greenbelt (SIMI_MN_G) and Mean Proximity Index of Greenbelt (PROX_MN_G). (3) When the values of other factors remain unchanged in the grid, some results will present: Increase of pland-C at 10% will cause LST value increasing 1.0℃. Increase of CA_W at 10  hm will bring about LST value decreasing 0.2℃. Increase of CPLAND_G) at 10% will give rise to LST value decreasing 1.0℃.  Increase of SIMI_MN_G and PROX_MN_G at 0.1 respectively will result in, LST value decreasing 0.09℃ and 0.08℃.
Key words：landscape pattern; thermal field; gridding; multi-factors; urban heat island effect