In order to fully utilize the capabilities of the densely distributed Global Navigation Satellite System (GNSS) network, this study has developed a comprehensive near real-time GNSS retrieval system for analyzing Zenith Total Delay (ZTD) and Horizontal Gradients (HG). This system is specifically applied to investigate the evolution characteristics of ZTD and HG during the heavy rainfall event on 7 May 2017 in Guangzhou. The analysis incorporates gridded precipitation products from the China Meteorological Administration and ERA5 reanalysis data, ensuring a robust comparison and validation of the results. The study’s findings reveal that the temporal variations in ZTD and HG had the potential to capture early indicators of precipitation, offering valuable insights into the atmospheric conditions preceding rainfall events. Specifically, the increase in ZTD values was observed to coincide with the onset of precipitation, suggesting that ZTD could serve as a precursor to heavy rainfall. Meanwhile, the horizontal gradients (HG) exhibited distinct directional patterns that correlated with the movement and intensity of rainfall, indicating that HG not only reflected the presence of precipitation but also provided information on its spatial distribution and dynamics. Furthermore, the vector characteristics of HG were found to reveal certain critical features related to rainfall, such as the direction of moisture convergence and the intensity of localised convective activity. Areas with high Horizontal Gradient (HG) delay tended to have higher Integrated Water Vapor (IWV) and were more likely to experience precipitation. In summary, the spatiotemporal characteristics of ZTD and HG, as derived from the GNSS network, offered significant potential for improving short-term precipitation forecasting, particularly for extreme weather events such as heavy rainfall. By providing early warning signals and detailed insights into the evolving atmospheric conditions, these GNSS-derived parameters served as a valuable auxiliary tool for meteorologists and researchers engaged in the nowcasting of severe weather. The integration of ZTD and HG data into existing forecasting models could have enhanced the accuracy and lead time of predictions, thereby contributing to better preparedness and risk mitigation efforts in regions prone to sudden and intense precipitation events.