To address the limitations of the traditional cross-scanning method in absolute calibration of weather radars using metal spheres, including insufficient spatial coverage, limited target acquisition efficiency, and echo underestimation in inter-range bins, this study proposes a sector scanning field calibration method. In this approach, standard metal spheres are suspended from UAVs, and a three-dimensional scanning volume around their theoretical positions is constructed to enable high-density echo sampling. By applying drive backlash correction, quadratic Gaussian surface fitting, and three-dimensional ellipsoid model inversion, key radar parameters can be retrieved. Experimental results show that the improved sector scanning method enhances automation, accuracy, and robustness in field environments and minor target drifts. The experiments were conducted under low-wind and low-clutter conditions. The average calibration error of antenna pointing is 0.08°, the average error of echo intensity calibration is 0.3 dB, the average beamwidth error is 0.07°, the range resolution is 6.6 m, and the average radial ranging error is 14 m. These results indicate that the proposed method can meet the main calibration requirements of weather radars in the present experiments.