Estimating monthly time series of streamflows at ungauged locations in the United States

Estimation of time series of monthly streamflows at ungauged locations is of paramount importance to many hydrologic projects and applications. This project uses leave-one-out experiments to assess the relative performance of a few traditional, regional, hydrostatistical prediction techniques at minimally-impacted streamflow gauges across the United States. This project considers four traditional flow-transfer techniques: drainage-area ratios (DA), standardization by mean (SM), maintenance of variance extension (MOVE) and the use of flow duration curves (QPPQ). Under idealized conditions, all methods significantly outperformed the drainage-area ratio (DA). However, when the flow-transfer techniques were combined with regional regression methods the relative performances were significantly degraded. A weighting scheme is introduced, combining the advantages of DA with the improved performance of MOVE or SM. This so-called weighted average (WAve) offers significant advantages over the traditional drainage area ratio techniques.