Abstract: This paper was concerned with monitoring of the farmland being returned to forests using Quickbird high resolution satellite data. And the returned farmland was investigated according to plan map supported by the GIS system. The relative data such as land patch area, land patch type and non-afforestation area were derived. An attempt was made to identify small trees within the returned farmland. In order to increase the identification rate, a NDVI transformation was performed to enhance the returned farmland vegetation.Unsupervised Classification base on an ISODATA decision rule of a fused image was used to create a map illustrated with different type of land patches. The area of different types of landuse and their percentage were calculated. The results showed that the detailed landcover such as boundary of crown of a tree, ribbing and compost could be extracted from the imagery.