Event-aware SWAT+ calibration for stormflows with monthly nutrient data in the Qingjiang River Basin

Storm-driven nonpoint-source (NPS) export in steep mountain basins is highly sensitive to runoff peaks, yet nutrient monitoring is typically monthly, limiting event-scale calibration. This study proposes an event-aware SWAT+ calibration framework for the Qingjiang River Basin (China) by coupling Hippopotamus Optimization (HO) with SUFI-2 uncertainty fitting. The objective retains monthly Nash-Sutcliffe efficiency (NSE) for discharge and nutrients and adds an event regularization term computed from daily discharge within objectively delineated runoff events (peak magnitude, time-to-peak, and recession slope). Using 2008-2018 for split-sample calibration/validation and 2020-2024 for an out-of-sample test without parameter retuning, HO-SUFI-2 improves monthly discharge NSE from 0.72 to 0.79 (calibration) and from 0.70 to 0.73 (validation) while tightening uncertainty (P-factor 83%→87%; R-factor 1.21→1.09). Peak constraints reduce median relative peak-magnitude error from 0.104 to 0.069 at Gaobazhou and from 0.114 to 0.069 at Changyang, and tighten peak-timing dispersion (|TPK| IQR 2→1 days at Gaobazhou; median |TPK| 1→0 day at Changyang), without degrading monthly nutrient skill (TP NSE 0.70→0.73). In 2020-2024, monthly discharge NSE remains 0.62-0.67 and TN/TP NSE ≥ 0.42 at both stations; remaining biases may reflect post-2020 management and reservoir operations not explicitly represented. The framework provides a practical way to constrain storm hydrograph dynamics under data limitations and to report uncertainty-aware diagnostics for storm-period risk screening in mountainous basins.