Abstract This study is motivated by the desire to enhance the understanding of subseasonal-to-seasonal (S2S) streamflow and flood predictability by addressing the challenges posed by S2S climate forcings (SCFs). Based on six numerical weather prediction (NWP) models and a calibrated hydrological model, we assess the predictability of precipitation, streamflow, and flooding across 24 river stations in the rainfall-dominated Pearl River Basin, South China. Various model configurations are conducted to address uncertainties related to initial hydrological conditions (IHCs) and SCFs, with ensemble streamflow prediction (ESP) serving as a benchmark for comparison. Key findings are: (1) NWP-driven hydrological forecasts significantly improve streamflow prediction accuracy, achieving Kling-Gupta Efficiency (KGE) values above 0.5 for up to 44 days, while precipitation forecasts reach KGE values above 0.2 for only 10 days; (2) NWP-based streamflow forecasts consistently outperform ESP, with KGE exceeding 0.6 for 21 days, compared to 5 days for ESP, and the Critical Success Index (CSI) for flood event detection remaining above 0.3 for three weeks, compared to one week for ESP; (3) Both SCFs and IHCs are essential for S2S streamflow predictability, with IHCs governing short-lead skill (especially in the first week), and improved SCFs providing a sustained source of predictability that extends skill to longer leads and, via more accurate state evolution between initializations, also strengthens IHCs at subsequent initializations. These results highlight the importance of considering both factors to achieve reliable long-term flood predictions.
Jiang et al. (Thu,) studied this question.