Global seismicity catalogs are sufficient for characterizing double seismic zones(DSZs)in subducting slab and facilitate to estimate layer separation without inconsistent uncertainties as local catalogs.Previous studies have shown the correlation between DSZs layer separation and plate age while correlation for those younger than ～60 Ma is suspicious.The lacking of DSZs with layer separation less than 10 km further makes it difficult to precisely estimate such correlation.Thus,we incorporate eight DSZs data determined through local seismicity into globally-determined dataset and reexamine such correlation.The best fitting results show that both a linear model and a square root of plate age can mathematically fit the layer separation well.However,it is difficult to distinguish these two models when plate age is greater than～20 Ma since their difference is less than 2 km.However,if extrapolation is possible,both models should provide physical information that DSZs will not form if there is no subducting lithosphere.As a result,the DSZs cannot be produced until the oceanic lithospheric age becomes greater than 0.9 Ma in the square root model while the linear model gives a misleading result.As such the square root model demonstrates the relationship physically better than the linear one,it still needs further test in the future with more available data,nevertheless,our study might also provide evidence for the suggestion that the plate age is a primary control factor of the DSZs geometry as well as the subducting process which disregards any local tectonic stresses. Global seismicity catalogs are sufficient for characterizing double seismic zones (DSZs) in subducting slab and facilitate to estimate layer separation without inconsistent uncertainties as local catalogs. Previous studies have shown the correlation between DSZs layer separation and plate age while correlation for those younger than ~ 60 Ma is suspicious.The lacking of DSZs with layer separation less than 10 km further makes it difficult to precisely estimate such correlation.Thus, we incorporated eight DSZs determined by local seismicity into globally-determined dataset and reexamine such correlation.The best fitting results show that both a linear model and a square root of plate age can mathematically fit the layer separation well.However, it is difficult to distinguish these two models when plate age is greater than ~20 Ma since their difference is less than 2 km. , if extrapolation is possible, both models should provide physical information that DSZs will not form if th ere is no subducting lithosphere. As a result, the DSZs can not be produced until the oceanic lithospheric age becomes greater than 0.9 Ma in the square root model while the linear model gives a misleading result. As such the square root model demonstrates the relationship physically better than the linear one, it still needs further test in the future with more available data, nevertheless, our study might also provide evidence for the suggestion that the plate age is a primary control factor of the DSZs geometry as well as the subducting process which disregards any local tectonic stresses.