Multiple DSG5 raindrop disdrometers are utilized to analyze the raindrop size distribution of the residual vortex of Typhoon Haikui in 2023 in the Pearl River Delta, and the performance of radar quantitative precipitation estimation (QPE) algorithms based on observed raindrop size distribution in the S/X-bands is evaluated. The results indicate that the impact of the residual vortex of Haikui on rainfall in the Pearl River Delta can be divided into two main stages. The first stage is primarily influenced by the convergence of the southwest monsoon and easterly airflow on the eastern side of the vortex itself, with a deep moisture convergence layer. The microphysical processes of rainfall are relatively uniform with higher number concentrations and relatively small particle sizes of raindrops. The second stage is characterized by severe rainfall caused by the convergence of southerly winds in the boundary layer on the outskirts of the residual vortex, with a shallower moisture convergence layer. The rainfall particles are more dispersed, with more rainfall events featuring large diameters and low number concentration of raindrops. Overall, medium-sized raindrops make a major contribution to the total rainfall amount, but as rainfall intensity increases, the contribution of extremely large raindrops to rainfall intensity at the second stage is significantly higher than that at the first stage. The dual-polarization parameters derived from raindrop size distribution of this process shows that, under the same rainfall intensity (R), horizontal reflectivity factor (ZH), specific differential phase (KDP), and differential reflectivity (ZDR) at the first stage are smaller than those at the second stage. The deviation of QPE algorithms in the S/X-bands calculated based on disdrometers data shows that, R(ZH) has relatively large biases in both the S-band and X-band, while the bias of R(ZH, ZDR) is relatively small in the S-band but increases significantly with increasing rainfall intensity in the X-band while R(KDP) has better performance in the X-band. R(KDP, ZDR) performs the best in both the S/X-band and is minimally affected by changes in raindrop size distribution, with little difference between the two stages.