Abstract: Jiuduansha, the mouth of the Yangtze River, is one of the most important wetlands in China and is a national nature reserve. Based on GF-2, Landsat and GoogleEarth remote sensing image data, this study interpreted the spatial distribution of Jiuduansha tidal creeks and landscape types. Ten tidal creek characteristic parameters including tidal creek length and curvature, and four landscape pattern indexes including patch area ratio and patch density were selected to explore the relationship between Jiuduansha tidal creek characteristic parameters and typical landscape patterns. The results showed that: The overall accuracy of U-Net convolutional neural network of extracting typical landscapes was 93.15%, and the Kappa coefficient was 0.94.The Jiuduansha area was less affected by human activities. From 2011 to 2021, the tidal channel network continued to develop, expand and become complicated. The length of the upper sand and upper sand tidal channels increased by 52.821 km and 133.467 km, respectively. Compared with the lower sand, the upper sand tidal channel development network demonstrated a better connectivity.In the absence of alien species and human disturbance, Phragmites australis was the dominant species of sand on Jiuduansha, and the area ratio of Phragmites australis patch was at a high value. Spartina alterniflora was the dominant species of sand under Jiuduansha, and the area ratio of Spartina alterniflora patch was at a high value.There existed a strong positive correlation between the length of tidal creek, the density of tidal creek, the curvature of tidal creek and the area ratio of reed patches (p≤0.05). The curvature of tidal creek was generally positively correlated with the degree of fragmentation of Scirpus mariqueter, and negatively correlated with the area ratio of Scirpus mariqueter patches.The curvature of the lower sand tidal channel was generally negatively correlated with the degree of reed fragmentation, and was strongly positively correlated with the area ratio of reed patches. The γ index was strongly negatively correlated with the area ratio of spartina alterniflora patches. The curvature of the tidal channel was significantly positively correlated with the area ratio of S.alterniflora patches (p≤0.01).By exploring the relationship between tidal creek development and wetland landscape pattern evolution, this research can provide a theoretical basis for the protection of Jiuduansha wetland.