Study on Transpiration Model for Fruit Tree Based on Generalized Regression Neural Network

In order to explore the water use characteristics of fruit tree and provide theoretical basis of effective and reasonable water conservation measures in Beijing. The transpiration of cherry was continuously monitored by heat pulse technology, and meteorological data were synchronously recorded. Because soil water contend (SWC), photosynthetically active radiation (PAR) and vapor pressure deficit (VPD) were closely related to transpiration, generalized regression neural network (GRNN) model for transpiration was constructed based those factors, and experiment verification showed that the model had a higher prediction accuracy.