Secondary Variables Selection and Data Pretreatment in Soft-Sensing for the Farmland Water Potential

In accordance with theory of water transfer in soil-plant-atmosphere continuum (SPAC), the moisture motion in SPAC is driven by water potential gradient. Therefore, continuous automatic measurement of water potential of SPAC and understanding of crop water stress is the most important premise of genuine precision irrigation (PI). In the view of the difficulty in auto-measurement of living crop water potential, a kind of special soft-sensing technique based on the microenvironment information around living crop is proposed. Secondary variables selection based on mechanism and sampled data analysis, outliers identifying and random noise filtering are important foundations of soft-sensing implementation.