Projection pursuit based correspondence factor analysis method of watershed non-point source pollution

In order to analyze the correlation between watershed non-point source pollutions and their sources, an improved correspondence factor analysis (named PP-CFA) was proposed based on accelerating genetic algorithm and projection pursuit method. The PP-CFA was applied to analyze the watershed non-point source pollutions of the Chaohu Lake. The results show that three kinds of correspondence relationship exist between land use and trophic salt loads of surface water runoff. The inputting characteristics of trophic salts of nitrogen and phosphorus of the rivers flowing into the Chaohu Lake can be classified as three kinds. Rational utilization of the land and effective control of trophic salt input are important approaches to reduce watershed non-point source pollutants. By using clustering method of scatter plot of one-dimension projection values, PP-CFA is more convenient to operation and with higher accuracy than other existing methods.