Estimating water consumption in Japanese plums

A project comparing three methods to determine water use finds that they provide similar results. By Jenny Underhill.

“We could estimate the water requirements of plums based on international work, but there wasn’t a dedicated study on plums in South Africa,” says Prof. Nebo Jovanovic.

Jovanovic is a researcher in the Environmental and Water Science programme at the University of the Western Cape. His work on water use in plums was funded by Hortgro and the Water Research Commission.

He measured water consumption in orchards of the early-maturing Fortune and the late-maturing African Delight on two farms in Wellington and Robertson.

One part of the study compared three methods to calculate evapotranspiration, which — after deducting effective rainfall — represents irrigation water requirements. “All three methods have advantages and disadvantages,” according to Jovanovic.

The three methods compared

The first method relies on eddy covariance micro-meteorological stations. “This is our reference method because it is the most direct method used to determine evapotranspiration,” says Jovanovic.

However, the stations each cost close to R1 million, require technical skill to use, and only collect data specific to a single orchard.

“In the second method, we used a model called HYDRUS 2D, which calculates water balances in soils under different conditions,” says Jovanovic.

To improve its accuracy, the model is calibrated against field data such as soil properties and tree-canopy shape. Drawbacks include the software licence cost — around R50 000 — and the time required to set the parameters and operate the model. Input parameters must be changed for every orchard.

The third method used remote-sensing technology. “Infrared sensors on satellites measure land-surface temperature to generate pixelated images,” explains Jovanovic. “Each pixel reflects a certain amount of light, and based on that, the amount of evapotranspiration is calculated with a complicated algorithm.”

This is the algorithm used by FruitLook, a paid service subsidised by the Western Cape Department of Agriculture.

“The benefit of FruitLook is that it gives large-scale data. For example, we would have to run the HYDRUS 2D model many times and have many micro-meteorological stations in different orchards if we want a broad picture,” says Jovanovic.

“However, satellite data don’t always provide the same level of resolution and detail as localised methods,” he cautions. “In addition, remote sensing has been found to be accurate under certain conditions and less accurate under others. To find out, you have to do ground measurements to validate the satellite data against a reference method.”

All three methods are accurate

When Jovanovic compared the three methods, he found only minor differences in the resulting seasonal water-use estimates.

For example, in one Robertson orchard, plum evapotranspiration values calculated from 1 September 2023 to 29 February 2024 were 751 mm with eddy covariance, 778 mm modelled with HYDRUS 2D, and 744 mm with FruitLook.

The results of the larger plum water-use study suggested annual water requirements of 7 300 m³ per hectare for African Delight in Wellington and 10 000 m³ per hectare for both African Delight and Fortune in Robertson. The final project report is due soon and will be published online by the Water Research Commission.