Modelling daily transmission losses in basin‐scale river system models under changing hydrological regimes

Abstract

AbstractWith rising concerns for water security and increasing interest in water resource development, accounting for river transmission losses in arid/semi‐arid region water budgets is a crucial yet challenging task. Transmission losses are usually confounded with many different processes and exacerbated by hydrologic and climatic non‐stationarity. A common deficiency in existing basin‐scale river system loss models is the poor representation of dynamic river losses into groundwater. To address this, a new conceptual model was developed to represent relevant river processes including rainfall to runoff transformation, river loss/gain to groundwater, river rainfall/evaporation and routing on a reach‐by‐reach basis. A critical process of the model is the exchange of water between the river and a groundwater store, called the river bed/bank store. The exchange to groundwater typically occurs with relatively high river volumes. Conversely, the store can discharge water back to the river when river volumes are relatively low. The model is designed to be transposable between different time‐periods, such as pre‐ and post‐development scenarios, assuming sufficient data exist during calibration. Using explicit Bayesian formulations, calibrations were performed against observed streamflow and remotely sensed evapotranspiration data. The new loss model was investigated using a test case in the Cooper Creek, Australia. The model performed overall better than a benchmark (flow vs. loss) model in a range of calibration/validation metrics. The new model provides river state and flux terms typically not available in basin‐scale models, and thus, is expected to be valuable during calibration/validation by allowing the use of alternative observed data types, for example, actual evapotranspiration or groundwater observations. Moreover, these extra terms could be very beneficial to water budget and ecological assessment studies.