Abstract: By using the Landsat TM/ETM⁺ thermal infrared remote sensing data of 1993, 2001, and 2011 to retrieve the ground temperature in the urban area of Lanzhou City, and by adopting objectoriented fractal net evolution approach (FNEA) to make image segmentation of the ground temperature, the urban heat island (UHI) elements were extracted. The G^* index spatial aggregation analysis was made to calculate the urban heat island ratio index (URI), and the landscape metrics were used at landscape level to quantify the changes of the spatial pattern of the UHI from the aspects of quantity, shape, and structure. The impervious surface distribution and vegetation coverage were extracted by a constrained linear spectral mixture model to explore the relationships of the impervious surface distribution and vegetation coverage with the UHI. In recent 20 years, the UHI effect in Lanzhou City was strengthened, with the URI increased by 1.4 times. The urban expansion had a spatiotemporal consistency with the UHI expansion. The patch number and density of the UHI landscape were increased, the patch shape and the whole landscape tended to be complex, the landscape became more fragmented, and the landscape connectivity was decreased. The conversion of the UHI mainly occurred from low temperature grade patch type to high temperature grade patch type. The heat island strength had a negative linear correlation with the urban vegetation coverage, and a positive logarithmic correlation with the urban impervious surface coverage.

Key words: Ningxia, urban human settlement, AHP, entropy method, coordination degree.