Site-specific irrigation enables the delivery of irrigation water where and when it is required

in the field. Commercially available hardware is available that can adjust the application

from centre pivots and lateral moves; however, wide adoption of these systems is limited

because of a lack of decision-support to determine irrigation application.

The irrigation control framework VARIwise was created to developed, simulate and compare

site-specific irrigation control strategies (McCarthy et al. 2010). This involves: (i) dividing

the field into smaller, controllable sub-areas named ‘cells’; (ii) assigning soil and plant

parameters to each cell; (iii) calibrating the corresponding crop model for each cell; and (iv)

executing a crop production model within in each cell. VARIwise has been used to

determine the optimal data input types and resolution for each control strategy in simulation

(McCarthy et al. 2011; McCarthy et al. 2012).

Two general types of adaptive control strategies have been implemented in VARIwise that

can be applied to irrigation: sensor- and model-based. Sensor-based control strategies use the

difference between a measured and target variable to update the irrigation application, whilst

model-based control strategies determine irrigation application that best achieves the desired

future crop performance as predicted by a calibrated crop production model. Two

implemented sensor-based strategies are Iterative Learning Control (ILC) and Iterative Hill Climbing Control (IHCC).

This project has demonstrated the implementation of adaptive control systems at commercial

cotton cropping sites. The control simulation framework VARIwise has been used for the

simulation and development of irrigation control strategies. The framework potentially

provides site-specific irrigation volumes and timing to be determined in autonomous

irrigation, either for uniform or variable-rate irrigation application.

Field evaluations of the control strategies were conducted on siphon and centre pivot

irrigation systems in Jondaryan, QLD utilising a weather, soil and plant sensors, control

strategies and irrigation control hardware. The siphon and centre pivot irrigation trials

produced yield improvements of 11% and 7% respectively, and water use reductions of 12%

and 4% respectively. Higher water reductions were achieved in surface irrigation systems

than overhead irrigation systems because of the larger volumes of irrigation water applied.

Adoption of these irrigation control systems would provide improved and automated

irrigation management and labour savings to the industry.

An on-the-go plant sensing system was developed to estimate plant density, plant height (for

leaf area index calculation), flower count (for square count calculation) and boll count, as

required to calibrate the industry crop production model OZCOT for the control strategy

operation. This sensing system was stand alone and platforms were developed that enabled

mounting to on-farm vehicles (e.g. moped) and irrigation machines. The centre pivot

irrigation trial indicated that plant data input was preferable to soil data input for model-based

irrigation control strategies.

There is limited control hardware currently commercially available for surface irrigation, and

commercial variable-rate solenoid-based irrigation adjustment hardware is available for

centre pivots and lateral moves. For the purposes of the field evaluations in this project, an

irrigation control hardware system was developed that was independent of the irrigation

system. This was based on adjusting the flow rate using a remotely controllable ball valve

and servomechanism and could be installed in-line with siphons and droppers on irrigation

machines.

Over 90% of the Australian irrigated cotton industry uses surface irrigation. Further

enhancements to the system would entail investigating the spatial resolution of irrigation

application adjustment for both surface and overhead irrigation systems. This would

potentially determine the data requirements of surface irrigation control systems, reduce the

sensors requirement (leading to reduced cost of the system) and increase the practicality and

uptake of the final system. In addition, the control strategies could be extended to consider

fertiliser application in surface irrigation systems, as the efficiency of the fertiliser application

is expected to be related to the efficiency of the irrigation application.