Seasonal dynamics of the relationship between landscape pattern and land surface temperature in a coastal city

The effects of land cover on LST have been found varying with season. Yet the seasonal dynamics of the relationship between metrics measured landscape pattern and LST has seldom been investigated. This study used a seasonal series of Landsat TM/ETM+ images in the year 2002 to examine the landscape pattern-LST relationship in Xiamen city. Landscape metrics and partial Pearson correlation were employed. The results showed that all the bivariate correlations between landscape metrics and LST changed with season to different extents. There were 7 consistently significant metrics across seasons for the year in Pearson correlation. All the configuration metrics lost their consistent significance in correlations with LST across seasons after partial Pearson correlation. Composition metrics of forest (PLAND3) was found to be the effective metric affecting LST consistently across seasons in the year. And the strongest bivariate correlation of PLAND3 happened in autumn for the study years.