To investigate the relationship between the full-bloom stage of rapeseed and the cumulative effective accumulated temperature at different lower threshold temperatures, utilizing phenological observations of rapeseed development stages and corresponding meteorological data from 2001 to 2024 in Xinghua, three developmental phases were defined: from green-up to full bloom, from stem elongation to full bloom, and from initial flowering to full bloom. For each phase, cumulative effective temperatures were calculated using lower threshold temperatures of 0  ℃, 5  ℃, and 10 ℃. Correlation analyses between these accumulated thermal metrics and the timing of full bloom were conducted, revealing that the cumulative effective temperature (with a base temperature of 5 ℃) from initial flowering to full bloom exhibited the strongest statistical relationship. This variable was therefore selected as the primary predictor in a random forest regression model developed to forecast the full-bloom date. The resulting model achieved a coefficient of determination (R²) of 0.928, with a root mean square error (RMSE) of 2.5 days and a mean absolute error (MAE) of 2.04 days. The model presents excellent predictive performance, satisfactory fitting effect and high prediction accuracy. This model can be directly used for the operational forecast of the rapeseed full-bloom stage in Xinghua, providing scientific support for determining the holding time of the “Thousand Mounds Rapeseed Flower Festival” and possessing favorable practical application value and promotion prospects.