Details

Autor(en) / Beteiligte

• Lachance-Cloutier, Simon
• Turcotte, Richard
• Cyr, Jean-François

Titel

Combining streamflow observations and hydrologic simulations for the retrospective estimation of daily streamflow for ungauged rivers in southern Quebec (Canada)

Ist Teil von

• Journal of hydrology (Amsterdam), 2017-07, Vol.550, p.294-306

Ort / Verlag

Elsevier B.V

Erscheinungsjahr

2017

Link zum Volltext

Quelle

Elsevier ScienceDirect Journals Complete

Beschreibungen/Notizen

• •Statistical interpolation (SI) combines observations from gauges and simulations.•SI compares favourably with other existing and widely used methods.•SI performance has a lower dependency on site location vis-à-vis the surrounding gauge stations.•SI offers potential research directions for improving estimates of daily streamflow. Retrospective estimation of daily streamflow for all rivers within a territory is of practical interest for sustainable and optimal water management. This implies, however, the availability of methods for providing accurate estimations of flow for ungauged rivers. This study compares the potential of statistical interpolation (SI)—a simple data assimilation technique that combines observations and simulations from hydrological modelling—with four other approaches: nearest neighbour, direct use of outputs from hydrological modelling, ordinary and topological kriging. Through subsampling cross-validation analyses based on the modified Kling-Gupta efficiency indicator, we show that SI compares favourably with these other approaches. While the performance of other methods depends on the configuration of the ungauged site in regards to the neighbouring reference sites, SI is less affected by these configurations. SI outperforms the other approaches particularly where the ungauged site is relatively distant from observation sites. In these cases, SI performance depends on the performance of the background model that relies on simulations of hydrological processes forced by precipitation and temperature observations. Our findings offer the potential for heightened performance estimates through an improvement of hydrological modelling and the use of more complex assimilation techniques for exploiting the model.