The urban form and land use data are crucial for urban simulations. Previous research usually treats the urban land as a homogenous identity, which is insufficient for modeling studies, especially for the climatic simulation. Besides, the land use data is not readily available in many developing countries such as China. Given this situation, the World Urban Database and Access Portal Tools (WUDAPT) can be used as an effective method to fill these gaps. Based on the land use classification system of the Local Climate Zone (LCZ) scheme, WUDAPT adopts free data sources (i.e. Landsat images and training samples from Google Earth) to generate LCZ maps for further numerical simulations. In this paper, the default WUDAPT method was introduced and modified. By adopting neighborhood information, the accuracy of land use classification can be significantly improved. Two LCZ maps of Dongguan city (2013 and 2017 respectively) were generated using this new method. These maps were used as input for the CA-Markov model to have future urban form and land use scenarios. Accuracy assessment was also conducted showing a satisfying simulation result. This paper is not only a meaningful attempt to expand the interdisciplinary application of WUDAPT products but also can serve as a reference for planning practices and decision-making process.