This paper examines wetland vegetation distribution based upon perennial water levels and the time and spatial characteristics of wetland vegetation in various phonological phases. Harmonic components are extracted to describe wetland vegetation index and the changing patterns are used by remote sensing data in a high time resolution. This paper defines one analysis unit by a period from 2000 to 2013 as one hydrological year (from September to the next September). Changing patterns of wetland vegetation during different analysis units are expressed by the period of harmonic component which has the maximum amplitude. Results show:(1) Area of wetland vegetation in Lake Poyang Natural Reserve is significantly influenced by hydrological characteristics. Compared to that in Nanji Natural Reserve, wetland vegetation distribution in Lake Poyang Natural Reserve is more sensitive to the changes in water level. (2) Area of the wetland vegetation within natural reserves is negatively Pearson correlated with water level in a hydrological year (at a 0.05 significance level). Impact of recession date is larger than that of next-year flooding date on distribution of wetland vegetation, while water level in dry season is not significantly Pearson correlated with the distribution of two-growing-period wetland vegetation. (3) The Pearson correlations between distribution of wetland vegetation on different elevations and water level in the two natural reserves show distinct correlations. In Lake Poyang Natural Reserve, wetland vegetation distribution from the elevations 12-13 m has the strongest correlation with water level in September, and the wetland vegetation distribution from the elevations 13-14 m has the strongest correlation with water level in October. In Nanji Natural Reserve, the wetland vegetation distributions on different elevations are significantly correlated with water levels in September and October. A harmonic analysis is conducive to further study on wetland hydrology relations based on multi-temporal remote sensing data.