The spatially-explicit nature of the model allows for robust representation of varying canopy architectures and their effect on berry temperature. The objective of this study was to accurately simulate the spatial and temporal grape berry temperature fluctuations from different vineyard designs, such that model predictions are robust to changes in vineyard configuration such as row spacing, trellis system, and row orientation.The 3D geometry of the ground, woody tissues, leaves, and grape berries were represented using a mesh of triangular and rectangular elements within the Helios 3D modeling framework. The procedural plant model generator in Helios allows the user to specify average and random geometric parameter values in order to create a given canopy geometry. Grape berries were represented in 3D as tessellated spheres composed of triangular elements, the ground surface was represented as a planar grid of rectangular elements, woody tissues were represented as a cylindrical mesh of triangular elements, and leaves were represented as a planar rectangle that is masked to the shape of a leaf using the transparency channel of a PNG image.When the terrain was flat, only one grapevine plant geometry was represented in the model, but with periodic boundary conditions applied in the horizontal directions which effectively yielded a horizontally infinite canopy. For inclined terrain, 7 rows of grapevine plants were represented in the slope-normal direction, and periodic boundary conditions were enforced in the slope-parallel direction .For triangular elements corresponding to grape berries, it was assumed that, large round pot since the density of lenticels after veraison is low, the latent heat flux was negligible.
In terms of Eq. 3.1, this assumption was implemented by setting gw = 0 for berry elements. It was also assumed that heat released from metabolic activity within the fruit is negligible, and thus no metabolic heat source term was included in Eq. 3.1. The formulation given by Eq. 3.1 also inherently assumes that the surface temperature at any point is representative of the entire temperature of the object . Furthermore, since the surface of the berry is discretized into many triangular elements that could have different temperatures, it was effectively assumed that each triangle corresponds to a “slice” of berry volume subtended by the triangle, which is at the same temperature of that triangular surface element, and responds to ambient changes with a time constant equal to that of the entire berry . Average berry element radiative properties for PAR and NIR bands are given in Table 3.1, which were measured for 3 replicates in the field . For rectangular elements comprising the ground surface, Eq. 3.1 was applied without the latent heat term , thus assuming that latent cooling due to water evaporation from the soil was minimal. The heat storage term in Eq. 3.1 was retained and used an assumed value of CprA of 30 kJ K−1 m−2. Clearly, the assumption of uniform temperature in the soil column when calculating the heat storage term is incorrect. However, the chosen value of CprA provided a reasonable temperature response time scale for the soil on the order of an hour and gave midday storage values between 20 and 30% of the net radiation flux, which is generally consistent with observations. Furthermore, since the focus of this work was on modeling the temperature of the berries and not the ground specifically, this approximation was deemed acceptable given that the ground temperature only has a secondary effect on modeled berry temperature in terms of long-wave emission.
Note also that convective heat transfer between the ground and berries was not modeled directly, but rather was specified using measured temperature and humidity near the berries.To validate the 3D model, field experiments were conducted in four Vitis vinifera L. cv. Cabernet Sauvignon vineyards from Sept 19th to Oct 10th during the 2018 and ´ 2019 seasons. Two study vineyards were located in Davis, CA and two others were located in Napa, CA , with each vineyard having a different trellis type. At the research site in Davis, the vines were on a flat terrain, and in Napa the vines were terraced along an east facing slope of approximately 30 degrees. The grapevines sampled in Davis were trained to the vertical shoot position trellis system and the Wye trellis system , while the Napa grapevines were trained to a Unilateral cordon and a Goblet system . The Unilateral vineyard was configured with one vine in each terrace level, while for Goblet there were two per level. The sampled vine in Goblet was adjacent to the terrace slope leading upward to the next row. The Napa Unilateral and Goblet grapevines were on the same slope at approximately the same elevation and row orientation. Thus it can be reasonably assumed that these two systems experienced nearly the same ambient conditions, making possible a comparison of the effects of canopy architecture. While the Davis VSP and Wye grapevines were in adjacent blocks with the same orientation, a direct comparison is not possible because measurements were collected in different years, and thus ambient weather conditions were different. For the four experiments, berry temperature was measured using 0.076 mm diameter type ‘E’ thermocouples . The thermocouples were inserted into the center of the berries in exposed clusters facing the east and west side of the vine, and at each side of the vine 4 thermocouples were placed in different berries within the cluster. Because berries could develop necrosis from being punctured by the thermocouple, thermocouples were relocated to adjacent exposed berries at least every 12 days to maintain relatively fresh conditions.In order to remove the effects of canopy-scale energy and momentum transfer and focus only on berry-scale transfer, ambient measurements were made near the clusters and used to force the model.
A weather station was installed at each study site immediately adjacent to the grapevines chosen for temperature measurements. The environmental variable inputs that were measured included incoming above-canopy photosynthetically active solar radiation , wind speed , relative humidity and air temperature . The incoming solar radiation was measured at a height of 3 m and was used to calibrate the incoming solar flux model in Helios as mentioned previously. The wind speed, relative humidity and air temperature were measured directly adjacent to the vine at the cluster height in order to estimate microclimatic conditions just outside of the berry boundary-layer. The sampling period for all weather data was 5 min. Specific humidity was estimated using the measured air temperature and relative humidity data, and since atmospheric pressure was not measured at the site, hourly average air pressure data from the National Oceanic and Atmospheric Administration local weather stations in Davis and Napa. To evaluate the model’s ability to simulate spatial and temporal fluctuations in grape berry temperature, experimental data measured on clear-sky days was used to drive the model and generate predicted berry temperatures. These temperatures were then separately averaged over east-facing and west-facing clusters and compared to average experimental values for the same exposure. The accuracy of the model was evaluated using the statistical error indices of normalized root mean square error , big round plant pot the coefficient of determination , and the index of agreement .An average characterization of weather conditions during the roughly 3-week period in which the weather stations were deployed is provided in Table 3.3. A more detailed graphical depiction of the measured air temperature, air relative humidity, wind speed, and of the calculated specific humidity time series data for the different experimental vineyard designs over the chosen validation period is shown in Fig. 3.2. During the 3-week period, the daily average air temperature was similar in VSP and Wye, with a wider average daily range of temperature in VSP compared to Wye . The daily average and average of maximum and minimum relative humidity were significantly higher in VSP compared to Wye, while the maximum wind speed measured in VSP was similar to Wye. Architectural differences between Wye and VSP were characterized by higher berry height, wider row spacing, and increased self-shading in Wye relative to VSP. Recalling that air temperature, humidity, and wind speed measurements were made at the height of the berry clusters, the higher berry height in Wye likely created daytime conditions of lower convective and radiative heat transfer from the warm ground to the fruiting zone, and overall greater turbulent mixing of warm, moist air out of the canopy. Specifically, during the validation day, the greater wind speed measured in Wye was likely responsible for the reduced air temperature and humidity at the measurement height compared to VSP . For Goblet and Unilateral, during the 3-week period, the daily average and average range of temperature were similar. However, the average maximum relative humidity in Goblet was greater compared to Unilateral and the maximum wind speed was significantly higher in Unilateral compared to Goblet . The architectural differences between Goblet and Unilateral were dominated by the higher berry height and wider row spacing in Unilateral relative to Goblet. Additionally, the close proximity of the Goblet vines to the adjacent terrace slope created an even larger ground view factor. During the validation day, similar to that observed in VSP relative to Wye, the proximity of the clusters to the ground and low wind speeds due to the tight row spacing in Goblet likely contributed to the increased air temperature and humidity fluctuations compared to Unilateral .Spatial and temporal variation in measured and simulated berry temperature is depicted graphically in Figs. 3.3 and 3.4, respectively. A sample visualization of the 3D distribution of the surface-air temperature difference for each vineyard is shown in Fig. 3.1.
During the night, all berry temperatures were near the ambient air temperature, and thus the spatial variability in berry temperature was small and did not vary noticeably among the vineyard designs. During daytime hours, berries in the shade tended to closely match the ambient air temperature, and could reach over 10◦C above ambient when in direct sunlight. The maximum berry temperature increase over air temperature measured in the field was 12.4◦C for VSP, 11.3◦C for Wye, 12.2◦C for Goblet and 14.0◦C for Unilateral . Besides, the closed canopy in Wye that limited berry sun exposure compared to VSP, it is likely that the greater wind speeds in Wye contributed to the enhanced sensible heat flux exchange and thus the reduced temperature as compared with VSP. For the Goblet and Unilateral vineyards, the east-facing slope and the ratio between plant height and plant spacing mainly determined the hours of berry exposure at the different positions, and therefore, the spatial berry temperature fluctuations. High berry temperatures tended to occur in berries in the west side of the vine during the afternoon when air temperature was warmer and there was exposure to direct sunlight . Measurements taken on berries of the west-facing clusters showed that the warmer afternoon temperature increased berry temperature up to 10◦C more than that of a similar east-facing cluster. As shown in Fig. 3.3 the highest temperatures on the west side occurred between 15:00 and 17:00 for VSP and Wye , and between 14:00 and 16:00 for Unilateral and Goblet . It is possible that in Goblet the lower wind speeds, along with proximity of berries to the slope, resulted in less canopy-scale turbulent mixing and a subsequent heating of within-canopy air near the berries compared to Unilateral.Modeled berry temperatures fluctuated rapidly with changes in absorbed radiation, sensible heat, and heat storage. Maximum values of simulated cluster absorbed radiation for the NW-SE orientations in VSP and Wye occurred about 3 hours before noon in the east side of the vine and 4 hours after noon in the west side . Horizontal canopy division in Wye increased self shading early and late in the day compared to VSP, which minimized exposure to direct sunlight. The widely spaced vines and high berry height in Unilateral favored greater berry light interception early in the morning compared to Goblet. In Goblet and Unilateral vineyard systems, the large difference in bunch exposure between the east and west side of the vines appeared to be dominated by the east-facing slope, which reduced the absorbed radiation in the west side of the vines compared to the east side. Vineyard geometry had a significant impact on the timing of cluster shading, primarily because of variation in row spacing relative to the plant height. While the absorbed radiation fluxes were positive during the day, the sensible heat fluxes tended to be negative during the day because the berries were warmer than the ambient air. Overall, the sensible heat losses were greater in the afternoon due to the greater difference between berry and air temperature during these hours.