Water Deficit Diagnosis Based On The Micro Change Of Stem Diameter And Leaf Temperature And Transpiration Estimation Of Greenhouse Grape

With the aggravation of water shortage,precision irrigation is the inevitable trend of greenhouse grape irrigation in the future.It is inevitable requirements for precision irrigation that accurate diagnosis of fruit tree water deficit and accurate estimation of transpiration.On the basis of previous studies,three irrigation treatments,high irrigation?W1?,medium irrigation?W2?and low irrigation?W3?,were set up to monitor the stem diameter,leaf temperature and stem flow of grape in greenhouse.The responses of stem diameter,leaf temperature and stem flow to environmental factors?soil moisture and meteorological factors?at different growth stages of grape in greenhouse were analyzed,and the greenhouse Portugal was discussed.In order to diagnose grape water deficit,the BP neural network models ANN^T?meteorological factors,soil moisture content in root zone?ANN^M?meteorological factors as input?,ANN^S?soil moisture content in root zone as input?were used to simulate the dynamic change of grape transpiration in greenhouse.The applicability of each BP neural network model to simulate grape transpiration in greenhouse was evaluated.Results:?1?The diurnal variation curve of relative variation of grape stalk diameter?RV?showed a giant toothed periodic change of about 24 hours.At different growth stages,the variation trend of grape stalk diameter with time is different,and there are differences between treatments.The RV curves of high irrigation water treatment showed a trend of fluctuation and decrease at the mature stage of coloring,while those of other growth stages showed a trend of fluctuation and growth,and the total increase of stem diameter was positive.The RV curves of stem diameter of medium irrigation water and low irrigation water treatment showed a trend of fluctuation and decrease at flowering stage and early stage of fruit expansion,while those of other periods showed a trend of fluctuation and decrease.The diurnal variation curve of grape leaf temperature in sunny days was"Ji"curve,while in cloudy and rainy days it was irregular.The leaf temperature of grape carved at the same time under different irrigation treatments was low irrigation>medium irrigation>high irrigation.The diurnal variation of the stemflow rate of grapes in sunny days showed a bimodal curve,while in cloudy days it showed multiple peaks.Compared with other treatments,high irrigation water treatment has higher stem flow rate.Leaf temperature was negatively correlated with air relative humidity?RH?,but positively correlated with solar radiation?Ra?,daily average temperature?Ta?and saturated water vapor pressure difference?VPD?,and the correlation was:Ta>VPD>RH>Ra.Stem flow rate was negatively correlated with RH,and positively correlated with other meteorological factors.The correlation degree was Ta>VPD>Ra>RH.?2?The daily maximum shrinkage?MDS?of grape stalk diameter was positively correlated with Ra,Ta and VPD,negatively correlated with RH,and correlated with VPD>RH>Ta>Ra.The daily increment of stem diameter?DI?was positively correlated with RH,but negatively correlated with other meteorological factors.The correlation was VPD>RH>Ta>Ra.Influenced by meteorological factors,the correlation between MDS,DI and soil water content in grape root zone was weak,which could not accurately diagnose water deficit in grape root zone.By introducing SI_MDS and SI_DI,the interference of meteorological factors can be weakened to enhance the diagnostic ability of micro-variation characteristics of stem diameter to soil moisture status.DI_MDS can be used to diagnose the soil moisture status of grape at different growth stages.SI_DI should be strengthened in the diagnosis of soil moisture status during color maturation.The empirical model of crop water stress index based on canopy Leaf-air Temperature difference can be used to diagnose the water status of grapes in sunny days of each growth period.The optimum time for determining soil moisture in the new growth period and flowering period should be about 15:00 and 14:00 in sunny days,respectively,and the fruit expanding period should be about 12:00,with reliable results.?3?Neural network model can be applied to the simulation of grape transpiration in greenhouse.The simulation accuracy and model stability of transpiration intensity in hourly scale of greenhouse grape under different irrigation treatments by ANN^T and ANN^M were higher than those by ANN^S.In high irrigation water treatment,ANN^M has smaller simulation error,its RMSE is 44.23 mL·h^-1,but its model stability is lower than ANN^T.In the treatment of medium water and low irrigation water,ANN^T has smaller simulation error and higher model stability.