In the context of rapid urbanization, the difference in cooling efficiency of different landscapes is the basis of understanding the influence mechanism underpinning landscape pattern on urban heat island effect. Existing studies on the heat island effect primarily use land use or land cover as the proxy for the source and sink landscape. This creates an information bias by ignoring the thermal characteristics of the landscape and the thermal relationship with adjacent landscapes. It is deemed necessary to explore the landscape classification method for the urban heat island effect and analyze differences in the cooling efficiency of landscapes with different characteristics. Based on "source-sink" landscape theory, this study constructs a landscape classification method for urban heat island using the western part of Shenzhen City as the study area. Sixteen urban landscape types were divided based on the extraction of surface features such as Normalized Difference Vegetation Index, Normalized Difference Moisture Index, and Impervious Surface Area and land surface temperature from Thematic Mapper remote sensing imagery in 2010. With the analysis of the temperature relationship under the equal distance gradient for landscapes with different surface features, the "source-sink" landscape of urban heat island was identified. On this basis, the composed landscape characteristics of the "source-sink" landscape were analyzed, and the cooling efficiency of landscape with area and temperature as influencing factors was calculated. Combined with the landscape pattern index, the cooling efficiency of separate areas and shapes of the sink landscape was analyzed. Five landscape types were identified as the urban heat island sink landscape. The area is approximately 533 km², accounting for 45.5% of the total study area. Results of cooling efficiency showed that the cooling efficiency of the "sink" landscape is 1.54, among which the contribution rate of green space and shadow landscape was higher. Based on the analysis of landscape patterns, it was found that the cooling efficiency both first increases and then decreases with the increase of the index value. The results can provide support for the shape and area setting of a cooling landscape in urban planning and urban green infrastructure design decision-making, mitigate urban heat island effect and increase the cooling efficiency at the landscape level.