Abstract: With the continuous deepening of urbanization in China, a large number of agricultural land is occupied by construction, which seriously threatens the food security and ecological environment. Therefore, it is an important foundation for protecting agricultural land to obtain the time and space information of construction land in time. In this paper, based on the long time series Landsat images, a method about extracting the information of agricultural land occupied by construction is proposed. Firstly, the original time series and interannual change rate series of various indexes are constructed. According to the interannual change rate series, the indexes that can distinguish the agricultural land occupied by the construction are selected. Then, the outlier detection algorithm is used to detect the outliers of the interannual change rate series to obtain the land use changed pixels and change time. Finally, according to the change time, the original time series is segmented three sub-series, and compared with the sub-series corresponding to the original time series of construction land and agricultural land to identify the pixel of agricultural land occupied by construction. The experimental result shows that the method can quickly obtain the change time and spatial location information of the agricultural land occupied by construction land. The overall accuracy of change time is 89.35%, and the kappa is 0.88. The overall accuracy of spatial position is 93.49%, and the kappa is 0.91.