An improved relative frequency method for flood season partitioning

Area of Science:

• Hydrology
• Water Resources Management
• Statistical Modeling

Background:

• Flood season partitioning is crucial for water resource management.
• Traditional relative frequency (RF) methods using Euclidean distance can lead to biased results due to overlooking statistical dependencies and varying variability.
• Objective flood season partitioning requires advanced statistical approaches.

Purpose of the Study:

• To improve the accuracy and reliability of flood season partitioning.
• To introduce an objective method for sub-seasonal flood risk assessment.
• To refine the relative frequency method by incorporating the Mahalanobis distance.

Main Methods:

• The study introduces the Mahalanobis distance to the relative frequency (RF) method for flood season partitioning.
• Mahalanobis distance adaptively normalizes the evaluation space, accounting for correlations and varying variability among sub-season flood frequencies.
• Monte Carlo simulations were used for rigorous validation of the proposed method.

Main Results:

• The Mahalanobis distance-based RF method provides a more precise characterization of flood frequency distribution.
• The improved method accurately identifies periods of high flood risk.
• Application to the Lhasa River Basin defined sub-seasons: Pre-flood (Jun 1-Jul 1), Main flood (Jul 2-Aug 12), and Post-flood (Aug 13-Sep 30).

Conclusions:

• The Mahalanobis distance offers a statistically robust improvement over Euclidean distance for flood season partitioning.
• This enhanced method improves water resource management by providing more accurate flood risk assessments.
• The proposed technique offers a reliable tool for hydrological analysis and water management strategies.