News and Media

Underestimating infiltration

by Jason Whyte

In studying the infiltration of water from rivers to the soil underneath, Flinders University postdoctoral fellow Jordi Batlle-Aguilar has a simple but novel concept — let’s start at the beginning. Jordi’s interest is in the variability of infiltration in dry river beds, particularly for intermittent streams that are typical in South Australia and regions of Western Australia and the Northern Territory.

‘Because the streambed is initially dry, water is rapidly sucked into the streambed at the onset of flow, but the infiltration rate decreases over time,’ says Jordi. His work focuses on discovering how important the initial rate of infiltration is in relation to the total volume of water infiltration. He explains that his data ‘is valuable for water management, because infiltration from streams is one of the main sources of groundwater in arid areas. If we better know how much water infiltrates we can better manage the volume of water we have available in an aquifer’.

Ring infiltrometers are commonly used to estimate infiltration from streams, and can give an average value of infiltration in 30 minutes or so, but they have their limitations. As the area measured is small, the value obtained is only representative of a few square centimetres where the measurement was taken, which Jordi calls `low spatial representability’. He explains the infiltrometers’ drawbacks: ‘One measured value can be very different from another a metre away. We want to evaluate infiltration for some length of a stream, but it is not appropriate to extrapolate one measured value, at the centimetre scale, to a total value for a whole stream’.

Obtaining infiltration values that more accurately describe the behaviour of a section of riverbed required putting the manual and small cylinders aside and thinking big.

In March 2011 Jordi’s team took their ideas to Pedler Creek, an intermittent stream in the Willunga

Basin, one of the sites funded through the NCRIS Groundwater Infrastructure fund. They isolated a 7m length of the dry stream bed using 2 m wide steel panels. The panels were dug into the stream banks and stream bed, allowing Jordi’s team to flood the region and control water height. Over 5 days, water stored in large water tanks was pumped into the isolated stream reach to flood it to three consecutively higher water levels, 20, 30 and 37 cm, holding each level constant for at least 24 hours. A gauge recorded the water height in the stream over time to ensure the water level was constant. By knowing the rate at which water was pumped from the water tanks into the stream to maintain the stream water level as constant, the researchers could deduce how much water was infiltrating over time. As it was necessary to decrease the pumping rate over time to keep the stream water height at the desired value, this indicated that the infiltration rate was becoming smaller as the experiment continued.

Based on field experimental data, a numerical model representing the experimental site was built. The model was used to estimate infiltration caused by six days of natural flow that occurred in July 2011. Results showed that the traditional method (which does not consider transient infiltration) would have underestimated total infiltration to the groundwater by up to 25%. These results highlight the importance of accounting for high transient infiltration rates at the onset of flow in streams of arid areas. As the transient effect can last for several days, this explains the large underestimation of infiltration by the traditional method.

The new information obtained from Jordi’s experiments shows that his project has made a significant contribution to the study of waterways.

Download a pdf version of this media release (100kb).