Predicting discharge using a low complexity machine learning model

• Enabling real time water quality management using collaborative networked farms.
• Discharge predictive model uses 3 simple field parameters and 12-month training data.
• M5 tree based proposed model, trained on real data, give R2 as 0.82 and RRMSE a 35.9%.
• 80% of the residuals for the predicted values fall within ±2 mm discharge depth/day error range.
• The proposed model gives comparable results when compared to contemporary research.

This paper reports on the validation of a simplified discharge prediction model that is suitable for implementation on a resourced constrained system such as a wireless sensor network, which will allow their operation to become more proactive rather than reactive. The data-driven model, utilising an M5 decision tree modelling technique, is validated using a 12-month training data set derived from published measured data. Daily runoff and drainage is predicted, and the results are compared with existing data-driven models developed in this domain. Results for the model give an R2 of 0.82 and Root Relative Mean Square Error (RRMSE) of 35.9%. 80% of the residuals for the predicted test values fall within a ±2 mm discharge depth/day error range. The main significance is that the proposed model gives comparable results with fewer samples and simpler parameters when compared to previous published research, which offers the potential for implementation in resource constrained monitoring and control systems.