Abstract: Farmland shelter forest plays an important role in protecting the northwest windy and sandy areas and provides a good growth environment for crops. To understand the spatial distribution characteristics of farmland shelterbelt, it is necessary to extract the shelter forest effectively. It is an important way to extract the farmland shelter forest by using the medium and low-resolution remote sensing data.In this study, farmland shelterbelt in Qingtongxia City, Ningxia as the research object. In the 2019 image, seven periods of Landsat8 OLI image data which can represent the characteristics of four quarters were selected to reconstruct the time series curve. On the basis of image preprocessing, water body index, vegetation characteristics and vegetation index were integrated, the phenological characteristics of vegetation were extracted, and other features in the study area were removed by mask, in order to achieve the extraction of farmland shelterbelt. At the same time, the accuracy of the extraction results is evaluated by using the field survey results and land use data.The results showed that, the overall accuracy of shelter forest extractionwas 85.16%, and 44 out of 50 sampling sites were accurately extracted with 88.00% accuracy. Farmland shelterbelts in Qingtongxia City are mostly presented in the form of forest belts, which are mainly distributed along drainage ditches.