Measurements of sap flow by the heat balance technique were carried out in six Granny Smith apple trees, in a six-years old and irrigated orchard in the SW of France, from July to October 1988. The orchard rows were 4m apart and 1 m between plants. Leaf area of the plants were determined in early August and early October, ranging from 4,7 to 11,3m2. Sap flow was computed each 20min and summed up for 24-hr totals.
Daily reference evapotranspiration (ETo) was computed by the Penman-Monteith equation:
ETo = [d /(d+Y)].(Rn-G)/L + [y /(d+Y)][900/(T+275)].U.D
being L the latent heat of vaporization; d the slope of the saturation vapour curve; y the psychrometric constant, and Y = y (1+ 0.33U); Rn the net radiation of the grass; G the heat flux in the soil (neglected in the daily scale); T the mean air temperature; U the mean wind speed at 2m above the vegetated surface; and D the atmospheric vapour deficit.
Meteorological data were collected in a weather station 12 km from the orchard, practically in the same altitude. The wind speed measured at 10m was reduced to 2m. For comparative purposes, the mean values of D were calculated in two ways:
Method 1: D = 0.5 [(eaTMAX + eaTMIN) - (edTMAX + edTMIN)]
being eaTMAX and eaTMIN the saturation vapour pressure at the maximum and minimum temperature and edTMAX and edTMIN the corresponding actual vapour pressure.
Method 2: mean of hourly mean values computed by the automatic wheater station.
For July, September and October, daily sap flow (SF) correlated linearly with the composite variable (ETo.A), being A the tree leaf area, that is, SF = K (ETo A). In an alternative way, the density of sap flow per leaf area (SFD) was correlacted with ETo. Periods in which the leaf water potential was lower than -2.0 MPa were considered as of occurrence of water deficit and discarded from the analysis. In all months occurred significative values of the coefficient of determination (r2), but computation of D by method 2 led to higher r22 values in all cases. The composite variable (ETo.A) also led to higher r2 values. For the three months the K values were about the same (0.38 to 0.41 for method 1; 0.47 to 0.54 for method 2). Pooling all the data, K was 0.39 for method 1, and 0.49 for method 2, independently of the variable used (ETo.A or ETo).