These practices include use of vegetative shoots to shade the fruit, using shade cloths to reduce berry radiative heating, spraying berries with a natural “sunscreen” such as a clay compound, water misting with sprinklers and, when slope permits, changing the row orientation to reduce or balance solar radiation exposure. Shade cloths, which are the focus of this work, reduce temperature by controlling the transmission of incoming radiation to the fruit zone,without completely blocking all incident radiation, which could lead to a reduction in grape quality and an increase in disease risk. Shade cloths can be used to cover the entire canopy and thus affect a large-scale change in vineyard microclimate, or they can be applied directly to the fruiting zone to localize their impacts to berry microclimate. Previous studies have reported that shade cloths are effective in reducing maximum berry temperatures, though they have used different methods for quantifying these effects. Mart´ınez-Luscher et al. ¨ reported that 40% black shade cloths covering the fruiting zone of the canopy reduced cluster temperature by 3.7◦C during the warmest time of the day, large round plant pots while Greer found that 70% shade cloths covering the grapevines reduced canopy temperature by an average of 4.6◦C throughout the day.
Similarly, thin shade cloths and plastic films covering the grapevines were shown by Rana et al. to reduce midday berry temperatures by 2◦C and 6◦C below air temperature, respectively. While it is clear that shade cloth can be effective in reducing overall berry temperature, a higher degree of control of berry temperature may be desired. In certain instances, it may be beneficial to reduce berry temperature by a defined margin to avoid negative trade-offs, while also balancing temperatures between opposing sides of the vine. However, many interacting variables are likely to influence the efficacy of shade cloth, such as row orientation, row spacing, trellis type, and topography. Due to the large number of important variables affecting berry temperature, it can be difficult to generalize the relatively small number of experimental results that are only able to explore a few variable combinations in order to predict the effect of shade cloth for a given vineyard system. Crop models provide the potential for generalizing the results of field experiments to predict the outcomes of proposed management strategies for a specific site or climate scenario. Such models could allow for optimization of the design or management of vineyards to mitigate elevated berry temperatures under current or future climates, given that a large number of simulations can be efficiently performed to cover a wide parameter space.
Previous work has developed models of spherical fruit temperature, including one study that simulated the effect of hail nets on apple temperature. In a recent study, a 3D model was developed and validated that accurately simulated the spatial and temporal temperature fluctuations of grape berries in vineyards with different climates, topographies, and trellises.However, the current version of this model is not able to represent the effects of shade cloth on canopy and berry temperature. The overall goal of this work was to enable model-based evaluation and optimization of strategies for grape berry temperature control using fruit zone shade cloth. With this goal in mind, specific objectives of this study were to: 1) develop a physically-based 3D model of grape berry temperature that incorporates the effect of shade cloth, 2) generate an experimental data set against which the model can be validated, and 3) quantify the interacting effects of different strategies for excessive berry temperature mitigation such as altered row orientation, row spacing, topography, and shade cloth density.The model of grape berry temperature was based on the 3D model described in Ponce de Leon´ and Bailey, and modified to include the effects of shade cloth. A brief description of the overall model is provided below, with a focus on novel additions associated with shade cloth. The model was developed within the Helios modeling framework and has been validated based on field measurements of berry temperature between veraison to harvest. The computational domain consists of 3D geometric elements that fully resolve the spatial structure of the plants, berries, and shade cloth .
The berries were represented by 3D tessellated spheres composed of triangular elements, the ground surface by a planar grid of rectangular elements, the woody tissues by a cylindrical mesh of triangular elements, and the leaves by planar rectangles masked to the shape of leaves using the transparency channel of a PNG image. The shade cloth was formed by rectangles masked to the shape of a grid using the transparency channel of a PNG image, where “holes” in the shade cloth were created based on a grid of transparent pixels.Radiation transfer was simulated using a backward-ray-tracing approach that ensures each geometric element is adequately sampled for both short-wave and long-wave radiation. The model launches a large number of rays from each geometric element to simulate the various modes of radiation transfer, including emission, reflection and transmission based on the radiative properties of each element. When a ray encounters a transparent pixel on an element masked by a PNG image, the ray continues with no interaction. This allows for an efficient fully-resolved representation of shade cloth with a large number of holes. To eliminate domain edge effects, periodic lateral boundaries can be enabled that effectively creates an infinitely repeating vine geometry in the horizontal. For berries, the latent flux term was assumed to be zero since latent cooling is typically considered negligible after veraison. For this study, heat released from metabolic activity ´ within the fruit was assumed to be low enough to be neglected. The heat transfer coefficient from fruit, ground, and leaf surfaces to the atmosphere, h, are specified as described in Ponce de Leon and Baile. Previous field tests suggested an important role of berry heat storage in accurately representing temperature dynamics, and laboratory and field tests have shown that the chosen values for Cp and rhoA result in berry dynamic temperature responses in close agreement with measurements.To evaluate the model accuracy, the time series of experimental measurements and simulated temperatures for berries furthest west were compared. Model agreement with field measurements was quantified using the error metrics normalized root mean squared error – normalized by the difference between maximum and minimum measured data, the coefficient of determination , plant pots round and the index of agreement. To analyze the modeled berry temperature results in the “simulation experiment”, we took the average temperature of five berries in each of five exterior clusters on both sides of the vine. For comparison purposes, the total canopy and cluster daily light interception was calculated by integrating the light interception fluxes over the day. The daily light interception was calculated per vine area for the canopy and per berry surface area for the clusters To compare the effect of shade cloth density on berry temperatures in the simulated vineyard designs, we used indices based on the duration of elevated temperature and the intensity . The 35◦C temperature threshold was chosen because it has been reported to correspond to the approximate onset of unfavorable effects of elevated temperature on grape composition. Furthermore, to identify conditions that effectively balance berry temperature between opposing sides of the vine, we constructed a conditional inference tree with the variables listed in Table 4.2. The conditional inference tree was constructed using the “party” package of R statistical software. The conditional inference tree is a tree-based classification algorithm that performs binary recursive partitioning of data into groups containing observations with similar values. Conditional inference trees are similar to decision trees, the main difference being that each node in the conditional inference trees uses a significance test of independence to select a predictor variable rather than selecting the predictor variable that maximizes the information measure. In the conditional inference tree, predictor variables are circled and ranked and the ones at the top have the highest correlation with the response variable.
The algorithm stops if the null hypothesis of independence is not rejected, however, for visualization purposes, the maximum depth of the decision tree was set to 4.The magnitude of the wind speed varied across the sampling period, but overall Unilateral tended to have greater wind speeds compared to Goblet . During the day, the wind in the fruiting zone tended to come from the east while at night the wind tended to come from the west for both trellis systems . Since the row orientation of Goblet and Unilateral was N-S and the vines were on an east-facing slope, the wind coming from the east tended to be of greater magnitude. For both Goblet and Unilateral, the relative humidity significantly decreased on 25 Sept. 2019, likely due to the increase in air temperature. Overall, during the sampling period, Unilateral tended to be slightly less humid compared to Goblet. This can be explained by the wider row spacing in Unilateral that enhanced mixing of the canopy air compared to Goblet .During the sampling period, average daily maximum temperatures of the west-facing berries under the shade cloths were 1.6◦C and 2.7◦C higher than the daily maximum air temperature in Goblet and Unilateral, respectively. The control west-facing berries with no shade cloth were up to 8.2◦C and 5.9◦C above the daily maximum air temperature in Goblet and Unilateral, respectively . The relative reduction in maximum berry temperatures due to the shade cloth in Goblet and Unilateral was consistent across the sampling period, which included a wide range of ambient air temperatures . Pulp temperatures of more than 40◦C were measured on 25 Sept. 2019 for unshaded berries in both trellis systems. For both Goblet and Unilateral, the berry temperature under the shade cloth was close to the air temperature in the morning, while air temperature differences between berries under the shade cloths increased in the afternoon. The peak in measured elevated temperatures happened at different times in Goblet and Unilateral , primarily due to the influence of the ratio between plant height and row spacing on berry exposure. In the evening, under low-light conditions, the berry temperatures were similar to the air temperatures, while at night, radiative cooling likely caused the pulp temperature to fall below the air temperatures. Under low light and at night, the temperature differences among treatments were small .The model was validated by using the experimental measurements to determine whether the modeled berry temperatures appropriately responded to the shade cloth relative to the control . The model reproduced the magnitude of the berry temperature increase over air temperature reasonably well for both Goblet and Unilateral. The largest source of error appeared to be due to the transition period when berry sun exposure began, where the time of the simulated maximum temperature increase tended to happen earlier than that of the measurement . This mismatch could be due to slight inaccuracies in determining the position of each berry and leaf.Figures 4.7, 4.8, and 4.6 summarize the results of the simulation experiment in which the effect of different row orientations, row spacing, and slope aspects on berry temperature were evaluated. Orienting vineyards N-S on a flat terrain allowed for uniform canopy and cluster exposure to solar radiation, since radiation is approximately symmetric about the N-S axis . However, the berry temperature on the west side of the vine significantly increased in the afternoon compared to the east side because hysteresis in air temperature causes asymmetry about solar noon. On average, berry temperature on the west side of the vines was greater than 35◦C for about 1-2.5 hours longer than the east side . Interestingly, although there was temporal asymmetry due to air temperature and temperature extremes, the net daily accumulation of berry growing degree hours was virtually identical between each side of the vine . On flat terrain, rows oriented NW-SE increased light interception and fruit overexposure in the afternoon and E-W reduced light interception and fruit overexposure in the afternoon . The high exposure to direct sunlight in NW-SE oriented rows resulted in simulated berry temperatures up to 7.8◦C higher on the SE side compared to the NW side . Compared to rows oriented NE-SW, rows oriented NW-SE had an additional 3 hours of canopy light interception above 200 W m−2 between 14:00 and 17:00 and berry temperatures greater than 35◦C for 2 additional hours .