Abstract:Aiming at the complicated variations of water quality and uncertain change mechanism of urban rivers and lakes in Guangdong, Hong Kong and Macao Great Bay Area, the “air-ground-water” integrated model of the Jinshan Lake Basin coupling different scales of land surface unit, river network and lake was established. The variation laws of water volume, COD mass concentration and TP mass concentration in 2017 were simulated, and the relative errors of the three indexes were within 15%. The simulation results show that the water quality in the Jinshan Lake Basin exceeds the standard seriously, and the water quality in the dry season is worse than that in the wet season. In December of the typical dry month, proportion of area with daily water quality inferior to class V standard is more than 96. 66%. There are 31 times of moderate and light rain in the year, and the water quality in the basin tends to decline in 1-3 d after moderate and light rain. There are 8 times of heavy rain in the year, after which the water quality is improved. The concentration of water quality index in the Jinshan Lake Basin has a significant positive correlation with cultivated land and rural residential area, and a significant negative correlation with forest land area, indicating that the water quality in areas with strong human activities is relatively poor.