Abstract: 　Objective To identify snail habitats and monitor the changes by combining remote sensing technique with the ground-based ecological data. . Methods. Of 100 survey sites selected throughout Poyang Lake, 75 were randomly identified from 574 land mass units: 50 were from snail habitats and 25 were from non-snail habitats based on a historical survey (1982-1984); 25 sites with habitats that did not have snails were also selected. Each site, covering .{10 000.} m2, was divided into a grid of 100 cells from which 20 cells were randomly selected. Snails, when present, were collected from a 4 m2 frame placed in the center of the selected cell. Satellite Landsat TM images were obtained for the same period as the ground survey data collected. Unsupervised classification was used to classify the images. Identified land-cover types were correlated with snail habitat. . Results . The sensitivity and specificity of classified snail habitat were 90.0%-95.6% and 61.1%-68.6%, respectively. Based on the classification, estimated snail habitat areas in Poyang Lake increased from 623.4 km2 in 1999 to 762.8 km2 in 2000. . Conclusion . Snail habitats are associated with grassland-covered areas. Classified images can be used to identify snail habitats, determine their areas, and monitor snail habitat changes caused by annual fluctuations of key environmental factors.

Key words: snail habitat, remote sensing, Poyang Lake