Abstract
Seawater intrusion in the Minjiang River estuary has gravely endangered the water security of the surrounding area in recent years. Previous studies mainly focused on exploring the mechanism of intrusion, but failed to provide a scheme for suppressing seawater intrusion. The three most relevant determinants to chlorine level, which represented the strength of seawater intrusion, were determined using Pearson correlation analysis as being the daily average discharge, daily maximum tidal range, and daily minimum tidal level. Considering the lower requirement of sample data and the ability to handle high-dimensional data, the random forest algorithm was used to construct a seawater intrusion suppression model and was combined with a genetic algorithm. The critical river discharge for suppressing estuary seawater intrusion determined using this model. The critical river discharge was found to gradually increase with the maximum tidal range, which in three different tide scenarios was 487 m(3)/s, 493 m(3)/s, and 531 m(3)/s. The practicable seawater intrusion suppression scheme was built up with three phases to make it easier to regulate upstream reservoirs. In the scheme, the initial reading of river discharge was 490 m(3)/s, and it rose to 650 m(3)/s over six days, from four days before the high tide's arrival to two days following it, and before falling down to 490 m(3)/s at the end. Verified with the 16 seawater intrusion events in the five dry years, this scheme could eliminate 75% of the seawater intrusion risk and effectively reduce the chlorine level for the remaining 25% of events.