Abstract: In GNSS slope monitoring, large height differences between reference and monitoring stations increase relative tropospheric delay errors, which severely limit the vertical accuracy of real time kinematic (RTK) positioning. To address this issue, an improved regional tropospheric model was developed. The model utilized high-precision zenith tropospheric delay (ZTD) model data from the reference and monitoring stations to establish a cubic polynomial relationship between ZTD and station height, while also accounting for seasonal variations. A coastal slope with significant elevation differences was selected as the study site to validate the model. Experimental results show that the proposed model in this paper significantly improves vertical positioning accuracy in the up component, with enhancements of approximately 15% and 8% compared with the Saastamoinen and Global Pressure and Temperature 3 (GPT3) models, respectively. These findings demonstrate that the model effectively mitigates tropospheric errors under large height-difference conditions and provides a practical solution for GNSS-based slope monitoring.