Identifying driving factors influencing the flood control performance of detention basin operations via artificial intelligence

Understanding driving factors for the flood control effect is essential for real-time flood detention basin management to balance flood peak reduction and inundation losses. Traditional driving factor identification is based on a single-factor correlation analysis, which does not reveal the mutual effect of exogenous inputs, or selecting the combination of driving factors with a trial and error method which is impractical under a lot of possible driving factors. To address these issues, this study proposes an artificial intelligence based framework. It integrates a previously calibrated hydrodynamic model that generates detention basin operation samples under different hydrological conditions and operating policies and a data-driven input variable selection algorithm that automatically selects the combination of driving factors. Take the effect of the Honghudong Detention Basin (HDB) operation on water level reduction at Hankou station in the Yangtze River basin as a case study. The results show that the detention basin operating policy is most relevant for the HDB operation effect, followed by the upstream hydrodynamic condition. This provides insights into the relevance of input variables to the detention basin operation effect and a deeper understanding of the underlying physical processes, which is meaningful for improving detention basin operation.