Although the modified Goldstein filter based on the local signal-to-noise (SNR) has been proved to be superior to the classical Goldstein and Baran filters with more comprehensive filter parameter, its adaptation is not always sufficient in the reduction of phase noise. In this work, the local SNR-based Goldstein filter is further developed with the improvements in the definition of the local SNR and the adaption of the filtering patch size. What’s more, for preventing the loss of the phase signal caused by the excessive filtering, an iteration filtering operation is also introduced in this new algorithm. To evaluate the performance of the proposed algorithm, both a simulated digital elevation model (DEM) interferogram and real SAR deformation interferogram spanning the L’ Aquila earthquake are carried out. The quantitative results from the simulated and real data reveal that up to 79.5% noises can be reduced by the new filter, indicating 9% 32% improvements over the previous local SNR-based Goldstein filter. This demonstrates that the new filter is not only equipped with sufficient adaption, but also can suppress the phase noise without the sacrifice of the phase signal. Although the modified Goldstein filter based on the local signal-to-noise (SNR) has been proved to be superior to the classical Goldstein and Baran filters with more comprehensive filter parameter, its adaptation is not always sufficient in the reduction of phase noise. In this work, the local SNR-based Goldstein filter is further developed with the improvements in the definition of the local SNR and the adaptation of the filtering patch size. What’s more, for preventing the loss of the phase signal caused by the excessive filtering, an iteration filtering operation is also introduced in this new algorithm. To evaluate the performance of the proposed algorithm, both a simulated digital elevation model (DEM) interferogram and real SAR deformation interferogram spanning the L ’Aquila earthquake are carried out. The quantitative results from the simulated and real data reveal that up to 79.5% noises can be reduced by the new filter, indicating 9% 32% improvements over the previous local SNR -based Goldstein filter. This demonstrates that the new filter is not only equipped with sufficient adaption, but also can suppress the phase noise without the sacrifice of the phase signal.