Soil moisture map construction by sequential data assimilation using an extended Kalman filter

摘要

Microwave remote sensors mounted on center pivot irrigation systems provide a feasible approach to retrieve soil water information, in the form of water content maps, for the implementation of closed-loop irrigation. Major challenges such as significant time delays in the soil water measurements, the inability of the sensors to provide soil water information in instances where the center pivot is stationary, and the inability of the sensors to provide soil water information in the root zone reduce the usability of these water content maps in the effective implementation of closed-loop irrigation. In this paper, we seek to address the aforementioned challenges and consequently describe a water content map construction procedure that is suitable for the implementation of closed-loop irrigation. Firstly, we use the cylindrical coordinate version of the Richards equation, also known as the field model, to model an agro-hydrological system that is equipped with a center pivot irrigation system. Secondly, soil water content observations retrieved from the microwave sensors are assimilated into the field model using the extended Kalman filter to form an information fusion system, which will provide frequent soil water estimates and predictions in the form of water content maps. The utility of the proposed information fusion system is first investigated with simulated microwave sensor measurements. The information fusion system is then applied to a real large-scale agriculture field where we demonstrate its ability to address the forgoing challenges. Three performance evaluation criteria are used to validate the soil water estimates and predictions provided by the proposed information fusion system.