Characteristics of urban heat island (UHI) source and sink areas in urban region of Shenyang

Based on the Landsat TM image on August 12, 2010, a 4 × 4 fishnet with 16 grids was built, and each grid was 3.5 km× 3.5 km. The changes of the LST (land surface temperature), urban heat island (UHI) source and sink areas and their corresponding percentages in the 16 grids were analyzed in order to find the optimum configuration of source and sink areas. The relationships between LST and remote sensing indices such as NDVI, NDBI and MNDWI, LST and LI (UHI intensity index), LI and remote sensing indices were discussed. The results indicated that the percentage of UHI source area is greater than that of UHI sink area in each grid. The 16 grids are divided into 5 groups according to the percentage of source area. There are 7 grids in the first group with greater 90.00%, 3 grids in the second group with 80.00%-89.99%, 2 grids in the third group with 70.00%-79.99%, 3 grids in the fourth group with 60.00%-69.99%, 1 grid in the fifth group with fewer 60.00%. The mean LST in the source area, sink area and the whole grid decreases with the decreasing of source area percentage, especially for LST in the sink area. The NDVI and NDBI values in each grid are fewer than 0.100 and greater than 0.100 of the first and second groups respectively, and their corresponding LI values are lower than 1.00. It suggests that they can enhance the urban heat island effect. The NDVI values in the two grids with 70.84% and 78.11% are greater than 0.100, the NDBI values are 0.030 and 0.106, and the corresponding LI values are 1.05 and 0.94. The effects of the third group on urban heat island are complex. The percentage of source area is fewer than 70% in the fourth and fifth groups, and their corresponding LI values are greater 1.00, which suggests that they can reduce the UHI effect. Analysis of relationships between indices showed that the relation of the LST is in a negative correlation with NDVI and is in a positive correlation with NDBI, while that of the LI is positive with NDVI and is in a negative correlation with NDBI. The MNDWI has no significant correlations with LST and LI. According to the above analysis, the study plot can reduce or enhance the UHI effect when the percentage of UHI source area is fewer or greater than the 70%, while it neither reduce nor enhance the UHI effect when the percentage of UHI source area is about 70%. Thus, the 70% of the UHI source area should be considered as a reference for urban planning in order to reduce the UHI effect.