Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.797343
Title: Development of a predictive irrigation scheduling framework
Author: Adeyemi, Olutobi
ISNI:       0000 0004 8503 5148
Awarding Body: Harper Adams University
Current Institution: Harper Adams University
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Precision irrigation scheduling is critical for improving irrigation efficiency. However, to realize a robust precision irrigation scheduling workflow, adaptive decision support systems need to be incorporated and enabled as part of the workflow. Furthermore, these adaptive systems should be developed to align with the three key requirements of precision irrigation: measurement, monitoring and management. The overall hypothesis of this research project was that data-driven models which are capable of predicting crop water requirements and the plant response to water supply can aid precision irrigation scheduling. There were three specific objectives which were formulated with the key requirements of precision irrigation in mind. The first objective focused on the need to ensure the availability of quality data from soil moisture sensors in order to realize robust irrigation scheduling decisions. The performance of three dielectric soil moisture sensors was evaluated under varying conditions of soil texture, bulk density, temperature and salinity. Results indicated that calibration equations developed in the laboratory improved the accuracy of these sensors for all conditions. The second objective focused on the development of data-driven dynamic models to aid the precision irrigation management of greenhouse cultivated lettuce plants. Dynamic models were developed for the prediction of the baseline temperatures and transpiration dynamics. Results indicated that the crop water stress index (CWSI) computed using the predicted baseline temperatures was significantly correlated with the theoretical CWSI and successfully distinguished the water status of lettuce plants receiving fractional irrigation amounts. The information contained in the residuals calculated from the measured and model predicted transpiration was exploited as a means of inferring the plant water status. This method successfully identified plants experiencing a shortage of water supply, achieving a sensitivity similar to stomatal conductance measurements. The third objective focused on the development of dynamic neural network models for the prediction of the volumetric soil water content (VWC). The application of the models for predictive irrigation scheduling was also explored. The models successfully generated accurate one-day-ahead predictions of the VWC with minimal input data pre-processing. Using model-based simulations of the potato growing season, it was demonstrated that a water-saving ranging between 20-46% can be achieved when these models are used in a predictive irrigation scheduling system. In conclusion, the study demonstrated the applicability of adaptive data-driven dynamic models for irrigation monitoring and management. The proposed adaptive models can be combined to realize a synergistic sustainable precision irrigation system.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.797343  DOI: Not available
Share: