Sampling site design plays a vital role in the survey of oncomelania hupensis, and directly affects the accuracy of statistical inference results. Currently, many problems, such as large samples demanded, poor representativeness and low efficiency in the traditional systematic sampling method in the marshland schistosomiasis epidemic area were met.  This research proposed a spatial sampling method of oncomelania survey based on the representative samples-SORS, in which the Fuzzy c-means clustering analysis of snail environmental co-variation was performed to explore the spatial distribution characteristics This method was applied to snail survey in Chayegang marshland of Henghu farm in the Poyang Lake region, error analysis of the collected representative samples was conducted, and a comparison study was made among the outcomes of the SORS, the random sampling method and the traditional systematic sampling method. The results showed that the SORS method proposed was obviously better than the random sampling method and the traditional systematic sampling method, the absolute errors of snail number are 0.440 7, 1.551 4 and 3.565 3, the relative errors of snail number are 1.048 1, 3.689 7 and 8.479 6, respectively. It implies the new snail sampling method could obtain higher estimation accuracy than the current snail survey method, which could provide new scientific basis for optimization of snail survey method in the marshland schistosomiasis epidemic area in China.

Key words：fuzzy c-means clustering; representative sampling points; sampling design; Oncomelania hupensis