Abstract:Measure of the heterogeneity of urban underlying surface is important to improve meso-scale numerical model and increase accuracy of urban weather forecast.Based on fractional Brownian motion theory and Landsat-TM satellite images,fractal dimensions of Beijing's typical underlying surface types and the entire urban region are estimated for 2011,which is used to measure the heterogeneity of Beijing's underlying surface structure.The results indicate that the fractal dimension can effectively measure the heterogeneity characteristics of Beijing's surface.The heterogeneity values for different typical ground object samples are ranked in descending order as business district,residential district,green field,crop,woodland,and water body.The regions inside 6th-ring road and suburban towns have significant heterogeneity,with fractional dimensions generally higher than 2.50.There is a low fractional dimension region in Beijing's center area within 2nd-ring road region.Fractional dimension of region between 2nd-ring road and 4th-ring road is highest and above 2.70,and it decreases outside of 5th-ring road.In a whole,the heterogeneity of urban underlying surface is distributed as firstly increase then decrease from city center to surrounding areas in Beijing.The exurb crop and woodland have low heterogeneity.For different underlying surface types,average fractional dimension of building,green land,unutilized land,crop,forest and water are 2.71,2.62,2.55,2.38,2.30 and 2.28,respectively.And the values of fractal dimensions show the quantitative measure of heterogeneity of various underlying surface types,which bears reference for quantitative description of complexities of urban boundary layer.