Abstract:Percent organic matter, total nitrogen, available phosphorus, and salinity were estimated in 176 topsoil samples. These samples were collected from a 9600 hm2 cotton field in northern Xinjiang Province. A fuzzy c-means clustering algorithm was used to assign these samples to management zones. The derivative of the objective function with respect to the fuzziness exponent, (δJ/δφ)c0.5, was used to determine the optimum fuzzy control parameters. The optimum number of the classes and the fuzziness exponent was 4 and 1.5, respectively. The average confusion index was 0.17 in all management zones. Thus, the overlapping of fuzzy classes at points was low and the spatial distribution of membership grades was unambiguous. To estimate the validity of zoning result, the general statistics analysis on the data was carried out. The zoning statistics showed that variation coefficients of soil properties decreased, while the means of the soil properties differed sharply between management zones. These results indicated that fuzzy c-means clustering algorithm can be used to delineate management zones by the optimum fuzzy control parameters. The management zones can then be used to help guide the rate of fertilizer application in an effort to manage soil nutrient levels more efficiently.