Abstract: For X-band dual linear polarimetric Doppler weather radar, the quality of differential propagation phase shift (φ_DP) is a key factor that affects radar application and thus data quality control is necessary. First of all, non-meteorological echoes need to be eliminated by using ρ_HV, texture of Z_DR and φ_DP; a new method, i.e., Linear Programming (LP), is then applied for quality control of φ_DP data. It is found that raw φ_DP datasets with 3, 5 and 7 smoothing points almost have no impact on LP results. Application of the LP method on data of X-band dual linear polarimetric Doppler weather networking radars in Beijing shows that the LP can effectively improve the data quality of φ_DP. Major effects are as follows. (1) φ_DP is monotonically increasing and specific differential propagation phase shift (K_DP) is non-negative after LP, which are consistent with physical attributes of φ_DP and K_DP; (2) the LP also can effectively eliminate the influence of the backward differential propagation phase shift (δ) caused by hail and other large particles, and thus improves the accuracy of K_DP calculated by φ_DP.