Figure 5: Effect of irrigation target soil matric potential (\(\psi_{\text{target}}\)) and irrigation threshold fraction (\(f_{th\text{res}h}\)) on total irrigation amount (Irr), irrigation starting date (Irr start), number of irrigation events (Irr events), irrigation frequency (Irr frequency), irrigation dose per event (Irr dose), soil moisture (SM) at 5, 20, and 50 cm depth, and yield. Shown are yearly average values for S09 and the year 2016.

3.1.2 Tree transpiration and fruit harvest

The comparison of measured sapflow with simulated transpiration expressed as water consumption per tree is presented in Figure 6. Observed sapflow varied significantly between different trees resulting in large inter-quartile ranges. The two model runs showed no difference in simulated tree transpiration despite the difference in irrigation amount and timing. In 2021, CLM5 showed higher values and a slight shift in the seasonal dynamic as a result of a too early onset of leaf development compared to the observed values (LAIsim in Figure 6). Simulated leaf duration and total transpiration agreed well with the measurements in 2022. Tree transpiration peaked in July with a measured monthly average of 12.5 (2021) and 20.2 L tree-1 day-1 (2022) and simulated values of 25.1 (2021) and 24.5 L tree-1day-1 (2022). The better agreement between simulated and observed values in 2022 followed a reinstallation that was performed after partial sensor failure and unreliable measurements that resulted in data gaps for the 2021 growing season. The 2021 data should therefore be handled with care when interpreting absolute values. Simulated maximum leaf area index (LAI) was reached in early July. Full canopy cover in the orchards occurred in the second half of June, so slightly earlier, based on visual inspection of the phenocam pictures (data not shown). Simulated leaf area and hence transpiration fell to zero by December, which broadly agreed with observed sapflow and leaf senescence deduced from the phenocam images.