Rock burst is one of the most severe dynamic hazards in the underground mining industry, and it is essential to establish a reliable method to achieve accurate prevention and control of it. This paper proposes a comprehensive spatial and temporal rockburst warning method based on the Mann-Kendall trend test (MKT) and spatially smoothed seismicity model, which can evaluate rockburst risk in real-time and quantitatively. The method uses real-time values of rockburst warning indicators as input, uses MKT to determine the temporal trend of each indicator and takes whether the trend conforms to the rockburst precursor law as the criterion for triggering the alarm. Then combined with the confusion matrix to evaluate the warning effectiveness of each indicator and optimize the indicators, and then obtain the real-time temporal rockburst hazard index Q of the mining area through data fusion. Finally, the spatial rockburst hazard index S contour map is drawn using the spatially smoothed seismicity model to identify local rockburst hazard zones and assist mine personnel in taking preventive and control measures of rockburst in time. The results show that the proposed model considers the precursor information of rockburst evolution from “time-space-strength” multi-dimensionally. The warning effectiveness of Q reaches 0.431, which is higher than any single indicator. The local rockburst hazard zone shown in the contour map of S is consistent with the actual location of strong tremors and rockburst events. The model has been successfully applied in the Wudong coal mine in Xinjiang, China, which can assist mine personnel in making efficient and accurate rockburst prevention decisions and effectively reduce their prevention costs.