Abstract
Drought and water shortage are major concerns in California and many parts of the world, and efficient water use is critical for growers. Water stress refers to the condition where the water demand exceeds the available water for a plant. The immediate goal of this research was to enhance existing water stress monitoring systems and develop a better irrigation scheduling system for pistachio trees in California. Currently, existing strategies for detecting water stress are either model-based or sensor-based, and each approach has limitations. In this project, we developed a data-driven model that combines model-based and sensor-based approaches and a system that takes advantage of both techniques. The test site was a pistachio orchard in Central California. During the growing season, extensive amounts of data were collected. Local environmental data such as ambient temperature, relative humidity, and pressure, were collected using sensors. Besides these, other collected data included multi-spectral aerial images, thermal images, sap flow, stem water potential data, and local. Aerial images were used to construct several vegetative indexes. A feature selection method was used to determine the most relevant input data. All the selected data was fed into different AI models. This paper discusses the results and shows the best approach for water stress detection in a pistachio orchard.
Learn more about the full research here.
