Prediction of Grape Sap Flow in a Greenhouse Based on Random Forest and Partial Least Squares Models

Understanding variations in sap flow rates and the environmental factors that influence is important for exploring grape water consumption patterns developing reasonable greenhouse irrigation schedules. Three levels were established this study: adequate (W1), moderate deficit (W2) (W3). Grape estimation models constructed using partial least squares (PLS) random forest (RF) algorithms, simulation accuracy stability of these evaluated. The results showed daily mean W2 W3 treatments 14.65 46.94% lower, respectively, than those W1 treatment, indicating average rate increased gradually with an increase amount within a certain range. Based on model error uncertainty analyses, RF had better different growth stages PLS did. coefficient determination Willmott’s index agreement exceeded 0.78 0.90, smaller root square d-factor (evaluation uncertainty) values did, higher prediction was more stable. relative importance predictors determined. Moreover, comprehensively reflected meteorological moisture content soil layers In summary, accurately simulated rates, which irrigation.