Abstract: Land cover has an important relationship with global ecological environment change. As the main way to get the large-scale land cover information, remote sensing classification technology plays an essential role in monitoring land cover change. However, due to the requirements of monitoring research on image time series and the characteristics of mixed spectrum presented by pixels, the accuracy of existing urban land cover classification methods still needs to be improved. In order to accurately assess the ecological effects of urban land cover changes, it is very important to find a simple and feasible remote sensing classification method with high classification accuracy. Based on Landsat 8 satellite remote sensing images, this research established a relatively representative urban land cover classification system. Then we set up a new urban land cover classification method by integrating the phenological characteristics of ground objects to the traditional supervised classification method. After that the improved classification method was applied to the land cover classification in Jintan district, and finally we statistically analyzed the change of area of each land cover type in Jintan district form 2016 to 2018. The results showed that compared with the traditional supervised classification method, the supervised classification method with phenological characteristics can effectively improve the classification accuracy of each land cover type. Moreover, with the implementation of the biodiversity protection policy of Jiangsu province, the area of each land cover type in Jintan district changed a lot from 2016 to 2018. On the whole, this research integrated the advantage of phenological characteristics to the process of land cover classification, which not only ensures the simplicity and feasibility of the classification method, but also effectively improves the classification accuracy, providing a reference for ecological analysis of land cover change.