Active distributed temperature sensing to assess surface water–groundwater interaction and river loss in braided river systems

The problem

Braided rivers recharge the aquifers beneath them, and river managers need to understand that exchange and how to measure it (Piégay et al., 2006). On the Canterbury Plains, leakage from the channels of the Waikirikiri Selwyn River is the main source of recharge to the aquifer beneath it (Larned et al., 2008). How and where such leakage happens is not well understood (Coluccio and Morgan, 2019; Rupp et al., 2008). It is also difficult to monitor. Channels and gravel bars shift continually (Gray and Harding, 2007; Reinfelds and Nanson, 1993), and floods and riverbed scour can damage equipment.

Heat is a widely used tracer of water exchange between rivers and groundwater (Anderson, 2005; Constantz, 2008). In active distributed temperature sensing (A-DTS), a cable is heated while optical fibres inside it record its temperature along its length. Moving water removes heat from the cable. The cable therefore warms less where flow is faster. Our goal in this study was to measure river loss across the width of a braided river channel.

What we did

At our site, the river loses water to the ground and its active bed is up to 60 m wide. Beneath it, gravels 3.5 to 4 m thick form a shallow, permeable aquifer that we call the braidplain aquifer.

We used horizontal directional drilling to bore two holes, each 100 m long, about 5 m below the riverbed and at right angles to the channel. We threaded a cable with optical fibres and copper heating wires through each hole and let the hole collapse onto it. We surveyed the first cable on 24 February 2021 and the second on 1 March. Each time we recorded temperature every 0.25 m for 5 minutes, then heated the cable at 15 W per metre for 20 minutes.

At each 0.25 m point we fitted an analytical solution for heat spreading from a line source in flowing groundwater (Zubair and Chaudhry, 1996; Bakker et al., 2015) to the measured warming. We excluded sections of cable near the banks that lay in unsaturated ground. The fit gives the thermal conductivity of the saturated gravel and the speed of groundwater moving across the cable. It does not give flow direction.

What we found

Water levels in shallow wells rose and fell with the river and remained below it. This confirms that water flowed from the river into the aquifer.

We estimate an average groundwater velocity of 12.0 m per day along the first cable (10.8 to 13.6 along its length) and 15.8 m per day along the second (12.4 to 19.4). Their mean is 13.9 m per day. Water moves only through the open pores, which make up 20% of the gravel's volume (a porosity of 0.2, from drill core). The volume crossing each square metre of ground is therefore the velocity multiplied by the porosity (our explanation). Assuming the flow is vertical, we estimate that river water entered the braidplain aquifer at 2.8 m³ through each square metre each day.

Because velocity changed little across the channel, we suggest that buried former channels have only a small effect on flow paths at our site. However, our 0.25 m measurement spacing is coarser than the roughly 1 to 2 cm scale that flow through large pores between gravels may have. The cables may not resolve flow paths that narrow.

On 24 February, hydrographers gauged river flow along a 333 m reach spanning both cables. The river lost 8,640 m³ per day over that reach, about 100 litres every second (our arithmetic). Because the well levels and the gauged loss both show water leaving the river, we are confident that the flow we measured beneath the channel is river loss. However, gauging gives the net loss over hundreds of metres of river, whereas the cables measure flow beneath a single crossing of the channel.

The velocity estimates depend on values we assumed. For the first survey, a distance between heating wire and fibre of 2.4 or 3.6 mm gave 9.6 and 14.4 m per day, against 12.0 for the base case of 3.0 mm. Porosities of 0.16 and 0.24 gave 14.4 and 6.3, and grain heat capacities 20% lower and higher gave 10.3 and 13.5. Moving the assumed start of heating from 240 to 252 s gave 15.4 m per day. It alone changed the fitted thermal conductivity, from 0.56 to 0.69 W per metre per °C. Future surveys will need precise cable specifications and more frequent readings early in the heating period.

Limitations

The method assumes that water crosses the cable in one direction. Our interpretation also assumes that the drillhole collapsed fully around the cable. The fitted thermal conductivities, 0.56 and 0.74 W per metre per °C on average, are lower than the 1.7 reported for greywacke bedrock with a porosity of 0.03 (Mielke et al., 2016). We consider them reasonable for gravels with a porosity of 0.2, although repacking of gravel and fines around the cable may also have lowered them.

The cables were costly to install, but they remained usable after a 1 in 50 year flood in June 2021, in which the river gauge and eight monitoring wells were damaged and lost. Two vertical cables have since been installed to 30 m depth beside the river. We hope surveys with them will show whether vertical or sideways flow dominates beneath the channel.

Glossary

A braided river flows in several channels that divide and rejoin around gravel bars.

An aquifer is a layer of rock or gravel that holds and transmits groundwater.

Recharge is water that enters an aquifer, here by leaking from the river.

River loss is the water a river loses to the ground along its course.

Fibre optic cable transmits light along glass fibres. Here the fibres are used to measure temperature along the cable.

Horizontal directional drilling bores a hole along a horizontal path beneath an obstacle such as a river.

A watt (W) is a unit of power. A heating rate of 15 W per metre means each metre of cable gives off 15 joules of heat each second.

Thermal conductivity measures how easily heat passes through a material, in watts per metre per degree Celsius.

Porosity is the fraction of a sediment's volume that is open space, here filled with water.

Unsaturated ground has pores that hold both air and water.

Heat capacity is the heat needed to warm one kilogram of a material by one degree Celsius.

An analytical solution is an exact formula, here for how temperature around a heated line changes with time and distance in flowing groundwater.

Greywacke is a hard, grey sandstone. The gravels at the site are made of it.

A 1 in 50 year flood has a 1 in 50 chance of being equalled or exceeded in any given year.

A hydrographer measures the flow and level of rivers.