How many metrics are required to identify the effects of the landscape pattern on land surface temperature?

Urban heat island (UHI) is a global issue as a result of urbanization. Land surface temperature (LST) is closely related to the thermal environment and energy budget of the earth surface, and is an important parameter in identifying UHI effects. Previous studies have proved the effects of landscape pattern on LST by using landscape metrics. However, the metrics used were inconsistent in number and type. Further, fewer studies tried to select representative metrics from the numerous metrics for LST indication. In this study we tried to explore the effects of landscape pattern on LST in Beijing by using the representative class level metrics selected through cluster analysis, factor analysis and regression. The results showed a composition metric such as PLAND (e.g. percentage of impervious surface in a landscape) alone explained about 56% of the landscape mean LST, whereas adding a configuration metric such as LSI (landscape shape index) or Gyrate_MN (mean gyration index) explained approximate another 6-12%. Adding more other configuration metrics does not improve the regression model performance more than 1%. The regression results also revealed that without targeted dependent variables, the factor analysis is of no use for choosing landscape metrics. These indicate that landscape composition and configuration both have effects on landscape mean LST, while composition is much more important than configuration, and that a combination of one composition metric with no more than four configuration metrics of impervious surface is sufficient for LST prediction. These results can help landscape ecologists in using landscape metrics and further help landscape planners to balance land cover in urban planning.