Abstract: With the pace of urban modernization, non-point source pollution is becoming increasingly serious to the urban rivers. According to the survey data, the concentrations of total phosphorus (TP), ammonium (NH₃-N) and suspended solids (SS) in Zhenjiang Ancient Canal were higher than the national standard. As a result, the water quality of the Ancient Canal has been seriously polluted, therefore it's urgent for us to take measures to monitor the water quality and build model to analyze the water quality of the Ancient Canal. Based on the monitoring data of rainfall, the content of contaminants in rainfall runoff in Zhenjiang Ancient Canal from 2009 to 2013, we established a regression equation between rainfall and runoff pollution, and then applied the Grey theory to predict future rainfall. Grey theory is the essential to make the irregular data variations into the regular data, through the way of accumulation, subtraction and so on to process, then using the differential equation fitting, and predict the future data by extension. What's more, the known original data is called white, the prediction data is called gray, and process is called bleaching or weakening of the randomness of the data sequence. The results reveal that the annual loads of TP, NH₃-N of rainfall runoff pollutants are expected to change with the rainfall in the canal, and the concentration of TP, NH₃-N and SS will change similarly from 2014 to 2020; when the rainfall reaches 1 381.2 mm in 2020, the annual load of TP, NH₃-N and SS will reach 217.15 t, 421.4 t, and 5 811.87 t, respectively and the concentrations of contaminants in rainfall runoff will be very large. Therefore, we should make green, clear the rubbish, take measures of the slope protection and a series of measures on both sides of the Ancient Canal to reduce the non-point source pollution produced by rain runoff, and then we can achieve good water quality of the river after taking action. Furthermore, the comfortable environment can provide a better life for the local people. Grey theory can weaken the fluctuation data sequence. regression analysis can establish the quantitative relationship to depend on each other between multiple variables. This paper combined the grey theory with regression analysis to propose the way to predict the load of rainfall runoff non-point source pollution, which can precisely forecasting the load of rainfall runoff non-point source pollution simply and effectively under the conditions of small sample, poor information and fluctuation data, and has great significance to promote.

Key words: rainfall runoff, Non-point source pollution, regression equations between rainfall and runoff pollution, Grey theory