Modelling the hydrological effects of in-channel leaky barriers for flood risk and natural flood management

A naturally formed and monitored log jam on the Logie Burn in Aberdeenshire during moderate flow conditions (0.46 m3/s; September 2017)

Background

In-stream ‘leaky barrier’ placement (e.g. log jams, flow restrictors) for reducing flood risk downstream has become more common. Accurate numerical modelling of their effects on water flow, is useful for predicting catchment or local scale changes in flood hazard. However, there are no commonly accepted approaches to representing leaky barriers within models. This work aims to produce guidance for the best approaches to modelling leaky barriers to reduce uncertainty.

Approaches to representing leaky barriers

Since leaky barrier type, porosity and river setting vary, there is no ‘one size fits all’ representation. Five types of representation used in hydrodynamic models were identified with the most common approach being the alteration of channel roughness and/or geometry.

The predicted hydraulic and hydrological effects in many studies should be treated with caution due to limited reported calibration, sensitivity testing and validation that has mainly been undertaken for low flow conditions. 

Provisional recommendations for representing leaky barriers in models are to:

1.       Use roughness values from the literature from similar rivers and of similar leaky barrier types as a starting point, or 

2.       Predict expected drag force using existing empirically based approaches and tailor the leaky barrier representation until modelled drag force equals the prediction. 

Field testing of modelling

Ongoing work is testing the validity of a 1D hydrodynamic model (HEC-RAS) to predict water surface elevations (WSE) and discharge of a monitored log jam (see image).  Initial results show that:

• The best predictions of WSE were obtained by representing the log jam with a high Manning’s n roughness coefficient (0.35-0.8)
• Using a weir representation resulted in poorer WSE predictions
• Discharge predictions were less certain than for WSE

Manning’s n values needed to be adjusted depending the inflow discharge.