Using Pattern Recognition Technique for Non-Uniform Field Sampling of Agricultural Nonpoint Source Pollution

As the variability of agricultural nonpoint source pollution in its distribution and generation processes, a field experiment on nitrate and phosphorus was carried out in the experimental base. Following the analysis of corresponding characteristics of the sampling measurements, a new method for non-uniform field sampling based on pattern recognition technique was presented, which could effectively avoid wasting resources and producing redundant information. In this method, pattern recognition technique was used to find out the variable boundary in the interesting region. After that a non-uniform sampling strategy was confirmed for the local regions with different variability. The results showed that this method could provide a more accurate dataset, whilst minimizing the sampling cost and simplifying the collecting procedure.