Abstract:

Quantifying the spatial differentiation of floods within a basin is crucial for effective flood control and disaster management. We analyzed daily runoff data from seven hydrological stations across different tributaries in the Poyang Lake Basin over the past 70 years by using the automatic peaks-over-threshold model and the master recession curve analysis method to extract three flood characteristics: total flood volume, peak discharge, and flood duration. We constructed a three-dimensional joint distribution for each hydrological station using the Vine Copula function and compared the flood patterns across stations by calculating joint return periods, concurrent return periods, and two conditional return periods. Our findings reveal the following: 1) The Log-Normal distribution best describes the marginal distribution of peak discharge, while the Gamma distribution most effectively fits total flood volume. 2) Gaussian Copula accurately represents the correlation between peak flow and total flood volume, whereas Gaussian and Student t Copulas are appropriate for the correlation between peak flow and duration under conditions of total flood volume. 3) The western Poyang Lake Basin is highly susceptible to catastrophic floods, characterized by higher total flood volume, peak discharge, and duration. 4) In contrast, the eastern basin can accumulate significant flood volumes quickly but does not experience prolonged floods. 5) The southern basin, however, may experience extreme peak flows given a substantial flood volume. These results offer valuable insights for enhancing flood warning systems and developing effective rated flood management strategies in the Poyang Lake Basin.

Key words: multi-dimensional joint distribution, Vine Copula model, flood characteristics, return period, Poyang Lake Basin