In recent years, with rapidly developing computational capabilities, high-resolution techniques have been continuously employed in the atmospheric models. Thus, researchers become capable of applying these high-resolution models to produce detailed meteorological scenarios, which empowers studies of urban-scale climatology which requires finer grid spacing. WRF ARW/Noah LSM/UCM model is often used in urban climate researches. However, the default input land surface data in urban areas is not enough precise, especially for the data describing in China. In this study, we developed a series of methods to produce the high-resolution urban land surface dataset which includes the land cover, vegetation coverage, urban morphology, artificial impervious area and anthropogenic heat data in Hong Kong-Shenzhen area. Moreover, a sensitivity test was conducted to examine the influence of urban land surface data accuracy on urban climate modeling quality. In short, the article proposes the methodological framework for producing the urban land surface dataset and presents how the urban land surface accuracy affects urban climate modeling quality.