Accurate and reliable high-resolution spatial precipitation data are crucial for hydrometeorology research. But most of the precipitation products have significant differences in terms of estimation accuracy owning to the influence of sensors, climate and terrain. Moreover, due to the neglect of the precipitation feature and the sparse distribution of gauge stations, the existing bias correction methods often have great uncertainties under different precipitation intensities. Thus, we developed a Daily Precipitation Bias Correction Approach Based on Feature Space Construction and Gauge-Satellite Fusion (BCFS). First, the precipitation feature space under different precipitation intensities was reconstructed, considering the attribute similarities of the spatial values, non-spatial values and trends. Then, the numerical relationships of correlated neighboring pixels were established taking account of these three similarities. Finally, the effective correction of the daily precipitation bias based on a small number of stations and a great number of pixels was achieved by the integration methods of variational mode decomposition, multivariate random forest regression model, and the spatial interpolation method. Using gauge station observations and the Climate Hazards Group Infrared Precipitation with Station data (CHIRPS) (1998-2019) and taking the Han River basin (China) as a case study, we quantitatively analyzed the accuracy of the bias correction results comparing the BCFS with the original CHIRPS precipitation estimations and the Wuhan University Satellite and Gauge precipitation Collaborated Correction method (WHU-SGCC). The results demonstrated the BCFS can effectively improve the estimation accuracy under different daily precipitation intensities. Therefore, the method is meaningful to make up for the deficiency of satellite-based estimations and provide high-precision daily precipitation for hydrometeorological and environmental monitoring and forecasting.