Erion Bwambale

CEA affiliates :

• WACWISA West African Centre for Water, Irrigation and Sustainable Agriculture, University for Development Studies, Tamale Ghana.
• RWESCK, University of Science and Technology, Kumasi, Ghana

Supervision :

Felik Abagale, WACWISA, UDS, fabagale@uds.ed.ugh

Co-supervision

Geophrey Anornu, RWESCK, KNUST, anoprof@hotmail.com

______________________________________________________

Title of the thesis:
Modeling and simulation of smart irrigation systems for more efficient water use.

• Google
• LinkedIn

Summary of the scientific project :
Global water scarcity continues to worsen, posing a challenge to sustainable agricultural production. The need to feed an ever-growing population has necessitated the adoption of innovative irrigation strategies, such as drip irrigation. However, conventional irrigation methods often fail to adapt to the real-time spatial and temporal dynamics of the soil, plant, and weather environment, resulting in over- or under-irrigation and, as a result, reduced water-use efficiency. With the increasing amount of data generated by advancements in sensor technology and improvements in computing power, there has been a growing trend to maximize crop yields per unit of water consumed in irrigated agriculture. In this thesis, a data-driven approach was used to model soil moisture dynamics using the System Identification Toolkit in MATLAB.

The sensor measurements for the agro-hydrological soil model were obtained from an open-loop experiment conducted on a tomato crop (Solanum lycopersicum L.) in an open field environment under drip irrigation. The soil moisture dynamics model was used to design a predictive model controller for intelligent irrigation programming. The predictive control algorithm was compared to a manual and open-loop irrigation control strategy to assess the resulting water savings and water use efficiency. The results of the data-driven modeling indicate that a state-space model represented soil moisture dynamics with better accuracy, with an estimated fit of 97.04% and a root mean squared error and final prediction error of 1.74 x 10-7 and 1.75 x 10-7 respectively compared to other model structures. The model's predictive control algorithm effectively captured soil moisture dynamics and accurately predicted future irrigation needs. The model-based irrigation programming approach resulted in 29% water savings and improved water use efficiency by up to 10.4 kg/m3 compared to the 7.1 kg/m3 and 5.6 kg/m3 open-loop irrigation strategies and the manual override irrigation strategy, respectively. These findings have important implications for farmers and stakeholders in the agricultural sector, as they offer practical solutions to address the current challenges posed by water scarcity.