Conducting geological hazard assessment and early warning in urban environmental design can help planners and designers better understand and respond to potential geological hazard risks. At present, the evaluation and warning of geological hazards in urban ecological design only rely on a single geological factor for analysis. They cannot integrate the many factors that cause geological hazards, forming a comprehensive and accurate geological hazard assessment method. The purpose of this article was to emphasize the importance of geological factors in urban environmental design, improve the accuracy of geological disaster assessment and early warning, and provide relevant countermeasures. This article found that the evaluation and warning of geological disasters had important guiding significance for professionals in the field of urban environmental design, which can help create safer, healthier, and more sustainable urban environments. This article used Back Propagation (BP) neural networks (NN) for risk assessment and early warning of geological disasters. With the powerful ability of NN to fuse and infer nonlinear factors, massive data is input into the NN for continuous training and evaluation, ensuring the safety of urban environmental design. This article considered the geological and geomorphological characteristics, including groundwater, slope stability, and flood control design. It fully utilizes and manages underground space, effectively utilizing limited land resources to enhance the sustainable development of cities. It also analyzed the use of engineering techniques to design urban environments, such as foundation treatment, earthquake prevention facilities, soil improvement, etc., to reduce the adverse effects of geological factors on the urban environment. To verify the effectiveness of the methods analyzed in the article, this article conducted a geological disaster assessment and early warning for towns 1–6 in H city. The highest accuracy rates for manually predicting 1–6 landslides, landslides, and mudslides in townships were 78.81%, 78.26%, and 79.44%, respectively. The highest accuracy rates for predicting 1–6 landslides, landslides, and mudslides in townships through BP NN were 83.29%, 84.25%, and 83.62%, respectively. This article highlighted the importance of geological disaster assessment and early warning for urban environmental design.