On 19 March 2020, Water NSW secured a conviction against two NSW irrigators (Defendants) for breaching water take conditions. The case provides some important guidance on the evidence required to secure a criminal conviction, and how the Courts deal with technical and scientific uncertainty.
Background
The Defendants owned and operated a cotton farm and jointly held a Water Access Licence (WAL), water supply works and water use approval. Under their WAL and water approvals, they were able to irrigate and take water from the Barwon River subject to the following condition:
‘Water must not be taken when the flow in the Darling River at the Bourke gauge (425003) is equal to or less than 4,894 ML/day’
Water NSW alleged that water had been taken by the Defendants between 22 June 2016 and 27 June 2016 when the flow in the water source was less than the volumetric limit in the condition, which is prohibited by s 91G(2) of the Water Management Act 2000 (NSW) (WMA).
The Defendants pleaded not guilty to the charges, and advanced a number of arguments, both legal and evidentiary, including:
(a) the transition of the water approvals was not effectively done at law when the WMA came into force, and so the Defendants were not bound by the condition;
(b) the separation of the charges in each summons was bad for duplicity as the incidents were not separate isolated offences;
(c) the Defendants were not properly notified of changes to the conditions in the letter from the Minister outlining the introduction of the new licensing regime; and
(d) the water level was not recorded from the Bourke Gauge on the days in question and therefore the prosecutor could not prove beyond reasonable doubt that the flow was less than 4,894ML/day.
Estimation of flow
The fourth issue raised the interesting evidentiary questions of how to prove beyond reasonable doubt that the flow rate at the Bourke Gauge was in fact less than 4,894 ML/day on the relevant days.
The difficulty for Water NSW was that it did not have direct evidence about the volumetric flow rate on the relevant days, despite water conditions routinely being expressed by reference to ML/day.
Instead, Water NSW calculated the volumetric flow rate of the river based on three variables, being:
(a) the height of the river, which was recorded by automatic sensor every 15 minutes;
(b) the cross-sectional area of the river through which the water is flowing, which is recorded by survey tools; and
(c) the velocity of the river, which was recorded by physical measurement (not automatically) at particular segments along the river cross-section.
For obvious reasons, Water NSW does not go to every gauge in its jurisdiction every day to accurately measure the three variables and calibrate in-situ recording equipment.
Water NSW instead relied on a ‘rating table’ methodology in accordance with the Australian Standard. Crudely, this involves plotting historic data for the three variables to calculate the volumetric daily flow under different conditions, and producing a ‘curve of mean fit’ to fill the gaps in data. The resultant rating table predicts the daily flow based on the height of the river (which is recorded each 15 minutes automatically).
The experts for both the Prosecution and the Defendants agreed that the rating table methodology involved a degree of uncertainty as rivers are hydraulic dynamic systems that change over time and also due to potential human and recording errors.
The arguments
The Defendants sought to challenge the Prosecution’s evidence on a number of grounds including that:
(a) the persons who collected the height and flow data were not identified or available for cross examination, and that their evidence would be inadmissible as opinion evidence;
(b) changes in the location of the Bourke gauge, uncertainty in the nature of the river channel, and infrequent calibration (approximately each 2 months) could result in inaccuracies in the ratings curve;
(c) the calculations and processes used to produce the rating table were flawed;
(d) the ratings curve did not accurately reflect the relationship between river height and flow rate at the Bourke gauge; and
(e) the height information for the river at the Bourke gauge during the relevant period was inaccurate.
The outcome
His Honour accepted the analysis from the Prosecution’s expert witness that accounting for the inherent error in the data, and any human error, an error of plus or minus 12.4% from the stated flow could be expected (with 95% confidence).
In accepting the Prosecution’s flow rate evidence, his Honour noted:
(a) the recording of height and flow data was admissible and could be relied upon as business records of Water NSW;
(b) the Rating Table was based on gauging’s taken by a range of instruments at different times and over a substantial time period (i.e. it was a large data set), adding to its strength as a reference;
(c) even if an individual measurement had been incorrect, the large data set reduces the importance of any one error;
(d) a margin of error may not help the Defendants’ position – the flow rate could have been lower than was measured and the true position worse for the Defendants; and
(e) finally, in this case, even if a 12.4% margin of error was taken into account in aid of the Defendants position (i.e. the flow was 12.4% higher than recorded), the flow rate was still below the requisite level.
Ultimately, the Judge rejected all of the Defendants arguments, and found that they were the owners of the land holding the approvals with conditions which limited the Defendant’s to taking water when the flow was above 4,894Ml/day and that they had breached the conditions in taking water during the specified period in June 2016.
The Defendants were both convicted of the offence as charged and are still awaiting the penalty. Each offence attracts a maximum penalty of $247,000 for an individual.
Conclusion
The case highlights the potential difficulty with using scientific and statistical methods in proving facts beyond reasonable doubt. If the Prosecution had simply relied on the ratings table, rather than independently scrutinizing the inputs and methodology prior to it being put to the Court, the Defendants may have generated enough uncertainty to avoid a conviction. The skeptical mindset with which the Prosecution approached its own data, and the rigorous analysis carried out for the case, coupled with the Court’s specialist jurisdiction likely made the difference.