To improve the accuracy of radar quantitative precipitation estimation (QPE), a high-precision dual-polarization radar QPE method is established, and its performance in operational application is evaluated. In this study, a nonspherical particle scattering model (i.e., T-matrix model) is used to simulate and calculate different polarization quantities on the basis of the data obtained using an LPA10 disdrometer. On the basis of the calculated results, the measured raindrop spectrum data are classified and fitted to optimize the precipitation estimation algorithm of CSU-HIDRO (Colorado State University-Hydrometeor Identification Rainfall Optimization). Two rainfall cases occurring in 2016 and 2017 in South China are selected to assess the performance of the modified algorithm (CSU-HIDRO_I). The R(Z_H) method PPS (WSR-88D Precipitation Processing System) and the CSU-HIDRO_I method are used to estimate the hourly precipitation. On the basis of different rainfall intensities and ranges (i.e., 20-60 km and 60-100 km) obtained by radar, the two precipitation estimation methods are evaluated. Moreover, the hourly precipitation estimated by radar is compared with that estimated by rain gauges. The main results are as follows: (1) The CSU-HIDRO_I method achieves good QPE results, and its estimation accuracy and stability are better than that of the R(Z_H) method. (2) The PPS method overestimates during light rainfall (R<2.5 mm/h) and underestimates during heavy rainfall and rainstorm (R>8 mm/h). By contrast, the CSU-HIDRO_I method can effectively reduce the underestimation of heavy rainfall and improve the estimation accuracy during light rainfall. Compared with the PPS method, the estimation deviation of the CSU-HIDRO_I method for light rainfall, moderate rainfall, heavy rainfall, and rainstorm is reduced by 38%, 24%, 17%, and 15%, respectively. (3) The PPS method is more sensitive to the distance from the radar during precipitation estimation than the other methods. Under the same rainfall intensity, the relative error at different distances fluctuates considerably. By contrast, the CSU-HIDRO_I method is less sensitive to the range from the radar than the other methods. Moreover, the variation of its relative error at different distances is smaller than that of the other methods.