[Influencing Mechanism of Precipitation Pollution Intensity in Qinhuai River Basin Based on Random Forest]

Analyzing the distribution characteristics of precipitation pollution intensity in the basin and identifying the main factors affecting the precipitation pollution intensity are the important basis for realizing the accurate management of diffused pollution. Based on the surface water quality data from four typical sections of the main stream of Qinhuai River Basin and rainfall data collected from 2021 to 2022, the distribution characteristics of precipitation pollution intensity in the basin were analyzed, and representative natural and social factors were selected to construct models of the precipitation pollution intensity of ammonia nitrogen （NH₄⁺-N）, permanganate index, and total phosphorus （TP） based on random forest algorithm. Additionally, the main driving factors of precipitation pollution intensity were identified, and the influencing mechanism was analyzed. The results showed that rainfall was the main driving factor for the water quality exceeding the standard in the Qinhuai River Basin, and the proportion of rainfall events within 3 days of when the water quality of the sections in the study area exceeded the standard was between 61.4% and 97.4%. The precipitation pollution intensity of NH₄⁺-N and TP showed a gradual increase from upstream to downstream. NH₄⁺-N was the primary pollutant in the midstream Qiqiaoweng section and the downstream Sanchahekou section, with the precipitation pollution intensity ranging from 0.12 to 2.98 and from 0.31 to 3.84, respectively. The precipitation pollution intensity of permanganate index in the upstream Yangqiao section was the largest, ranging from 0.41 to 1.35, and the precipitation pollution intensity of the Jiangjundadao section, which is located on the channel of diverted water, was at a lower level compared with that of the other three sections. Water diversion was an important influencing factor for the reduction of precipitation pollution intensity of NH₄⁺-N and permanganate index （P<0.01）； the proportion of built-up area was significantly positively correlated with the precipitation pollution intensity of NH₄⁺-N and TP； the proportion of cultivated area was significantly positively correlated with the precipitation pollution intensity of permanganate index, and the proportion of forest land area was significantly negatively correlated with the precipitation pollution intensity of these three indicators. Under the conditions of light rain, moderate rain, and heavy rain, the rainfall was significantly positively correlated with the precipitation pollution intensity of each indicator, and the precipitation pollution intensity remained high under the condition of torrential rain. The post-rain pollution of NH₄⁺-N and TP mainly originated from the middle and lower reaches of the study area, which was significantly affected by regional characteristics such as the percentage of built-up area and water diversion （P<0.01）, and the post-rain pollution of permanganate index mainly originated from the upstream area with a wide distribution of cultivated land area and was significantly affected by flow （P<0.01）.