Abstract: The difference between reflectivity factors observed by different radars observing the same target affects the performance of the multi-radar mosaic. For X-band radars without beam blocking, this difference is mainly caused by calibration deviation and signal attenuation due to rain and wet radome. To correct these errors, a new correction algorithm for network phased array weather radars is designed. In this new algorithm, we first regarded the attenuation due to water film on the radome as part of the calibration deviation, and used the network attenuation correction algorithm to make an initial attenuation correction. Then, an observation deviation function of the radar network is designed, which used the gradient descent method to solve the calibration deviation issue between different radars. Finally, a secondary attenuation correction, which used the network attenuation correction algorithm, is applied to the original reflectivity factor after calibration deviation correction. In this study, a network consisting of seven X-band phased array weather radars in the Foshan city of Guangdong province is used to verify the new correction algorithm. For a rainfall event associated with cold air, a correlation coefficient of 0.53 between the reflectivity factor corrected with the traditional path-integrated attenuation (PIA) algorithm and the reflectivity factor observed by an S-band weather radar in Guangzhou was obtained, with a root mean square error of 9.0 dB. While those between the new correction algorithm and the S-band radar are 0.64 and 8.4 dB, respectively, with both being better than PIA. A similar performance was also achieved in a typhoon outer rain-band rainfall event analysis.