4.1 Scenario a: closed greenhouse, lychee trees, March 2020
During February and early March 2020, the EC system was installed in a closed plastic greenhouse, with over 2.5 m tall lychee trees, irrigated with geothermal water. Recent studies (Stern 2018; Elimelech 2022) have shown that heating lychee trees immediately after the accumulation of "chilling hours” accelerates growth of their blossoms and results in early blooming and fruit that ripen in early May, about 1.5 months before regular commercial orchards.
It is obvious that the assumptions for flux calculations using an EC system are not valid in such a closed and limited environment. However, the instruments measure parameters that might be used without any calculations, such as CO2 and water vapor mixing ratio/concentration, air temperature, air pressure, humidity, etc. Thus, those parameters represent the real micrometeorological conditions in the greenhouse.
Figure 1 concentrates on 3 consecutive days during that period. Figure 1a shows fluxes that are low, and both CO2 and H2O fluxes exhibit strong positive and negative fluctuations that do not seem to be coherent. Although a general trend can be observed in the water vapor flux (positive flux -away from the plants with considerable fluctuations during the day and flux close to zero at night), those results cannot be considered accurate due to the lack of fulfillment of the EC calculation assumptions However, Figure 1b shows the results of direct measurements inside (temperature, and CO2 and water vapor mole fractions) and outside (temperature and global radiation) the greenhouse.
On March 1, the outside temperature ranged from 7–15oC; global radiation increased during the morning to ~250 W m-2 but in the early afternoon, cloudiness increased and global radiation decreased to ~150 W m-2. The temperature inside the greenhouse followed the general pattern of the radiation, reaching close to 30oC in the late morning, but when the cloudiness increased, the greenhouse cooled down by 10oC in less than 90 min. As the global radiation increased again, the temperature rose to 31oC in about 1 h (at 1530 h), and then steeply decreased when evening approached. A similar trend can be observed on March 2. On March 3, there was no cloudiness or daytime reduction in global radiation, so the inside temperature reached values of >30oC, and remained high, even after sundown (1700 h).
A very interesting correlation can be seen between the greenhouse temperature and the H2O mole fraction. At first sight, it can be assumed that this indicates constant relative humidity, since saturation water vapor pressure increases with temperature, but a detailed analysis (see Supplemental Material "Closed Greenhouse 03 2020" Excel file) shows that the relative humidity decreased during the high-temperature periods. The H2O mixing ratio and temperature correlated with a very good fit along the entire 3-day period (correlation = +0.910). Indeed, many researchers have reported a good correlation between ecosystem respiration and temperature (Liang and Wang 2020). The only period of misfit was on March 3 at around noon, when the temperature remained high, but the H2O mole fraction decreased from >35 to ~20 mmol mol-1. This decrease was accompanied by a clear increase in the CO2 mixing ratio. This combination of effects is generally ascribed to stomatal closing caused by heat stress caused by the high temperatures. Even when considering the short period of closed stomata due to heat stress, a negative correlation between water vapor and CO2 mole fraction was obvious during the entire 3-day period (correlation = -0.846), very clearly exhibiting the relationship expressed in Equation (2): when no "true" photosynthesis occurs, CO2 increases and H2O decreases, and vice versa during the day with open stomata.
As mentioned above and seen in Figure 1a, the behavior of the H2O and CO2 fluxes in a closed greenhouse appeared random and chaotic, but calculated correlation still yielded a relatively high negative value (-0.801).
To summarize, in the closed greenhouse, even though it is not hermetically sealed, the diffusion from "outside" to "inside" is relatively slow, and during high photosynthetic activity, the CO2 mixing ratio decreases to values that are almost half of the known average value of about 400 mmol mol air-1, approaching values of ~200 mmol CO2 mol air-1, close to those considered the lowest threshold for active photosynthesis (Moss 1962; Moore 2015). We could not find such low values anywhere in the literature. To ensure that these values were not due to a malfunction in the equipment, they were confirmed with a Rotronic CP11 handheld CO2, humidity and temperature measuring instrument. On the other hand, during the night, CO2 reaches values >600 mmol mol-1. A similar but opposite behavior is observed for H2O mixing ratio: values go down to ~10 mmol mol-1 at night or when the stomata are closed and increase to >35 mmol mol-1 during periods of high "true" photosynthesis.
4.2 Scenario b: open greenhouse, lychee trees, December 2021
On 12 March 2020, a strong easterly windstorm with gusts >100 km h-1 partly destroyed the greenhouse roof. The greenhouse was reconstructed, and an improved automatic system that opens or closes curtains depending on the temperature was introduced. Scenario b (Figure 2) shows measurements performed on 27–29 December 2021—before the automatic system was operational; the side curtains of the greenhouse were constantly open (day and night), and the only coverage was from the roof.
Considering that the main assumptions for the calculation of fluxes using an EC system were again not valid—due to the roof that hinders vertical turbulence and other greenhouses in the surrounding area, the fluxes in Figure 2a showed strong fluctuations; nevertheless the general trend observed in March 2020 (Figure 1a), where water vapor flux is positive during the day and close to zero at night, was maintained. However, the correlation with CO2 flux was very low (-0.136).
Figure 2b shows the direct measurements, which differ considerably from the measurements in March 2020. Outside temperatures ranged between 2.8 and 21.6oC. Inside temperatures, which in Figure 1 differed considerably from the outside values, follow them almost completely in Figure 2. This can be ascribed to the fact that the greenhouse was completely open to the outside environment. Global radiation increased during the morning, and some noon cloudiness was observed on December 28 with no significant influence on the temperature.
The direct and strong correlation between the greenhouse temperature and the H2O mole fraction observed in Figure 1b was not apparent here (correlation = +0.381), and we ascribe this to the open sides that allow rapid gas exchange between the interior of the greenhouse and the outside. Similarly, the steep decrease in CO2 mixing ratio observed in the closed greenhouse was not observed here, and the negative correlation between water vapor and CO2 mole fractions was not very significant (-0.394). Thus, photosynthesis cannot be monitored under these conditions by either fluxes or concentrations.
4.3 Scenario c: closed greenhouse with automated mechanical curtains, lychee trees, January 2022
The automatic system was adjusted to open the curtains during the day when the temperature inside increases above 30oC, and close them when it decreases below 26oC. During the winter at night, to avoid high heat stress on the trees, the curtains are left open if the temperature increases above 20oC (Menashe Levy, personal communication, 2022). Scenario c (Figure 3) shows measurements performed in the greenhouse on 13–15 January 2022, with the automatic curtain system fully operational.
As in the previous scenarios, the main assumptions for EC calculation of fluxes were not valid. Indeed, the fluxes shown in Figure 3a seem to show "random" fluctuations, as in the previous scenarios. Interestingly, unlike the open greenhouse scenario, the negative correlation between both fluxes was relatively high (-0.830) and similar to the value observed in scenario a (closed greenhouse); thus, as CO2 flux decreases, H2O flux increases, and vice versa.
Figure 3b differs from the open greenhouse (Figure 2b) and is relatively similar to the trends for the closed greenhouse (Figure 1b). During this period, there was a relatively warm winter days, and outside temperatures ranged between 12 and 29oC. During the entire period, global radiation increased during the morning but was influenced by some cloudiness. The inside temperature, similar to Figure 1b, differed considerably from the outside values, and from Figure 2b. However, an interesting effect could be observed, ascribed to the mechanical curtains’ action: while, for example, on 3 March 2020 (Figure 1b), the inside temperatures increased to values above 30oC and remained high for several hours, the "cooling" effect of the open curtains can be clearly observed at noon on all 3 days during January 2022: above a certain temperature, a sudden decrease in this parameter is observed, accompanied by an increase in CO2 and decrease in H2O . Furthermore, we note that on January 15, the inside temperatures increased again in the late afternoon (1630 h) after steeply decreasing at noon, completely unrelated to the temperature outside. We can explain this as follows: at noon, the temperature reached the higher limit, the curtains opened, and cooler air cooled the greenhouse. In the late afternoon (but still daylight), temperatures reached the lower limit, the curtains closed, and the greenhouse heated up again. This entire effect can also be clearly observed in the fluctuations of H2O and CO2 mixing ratios.
The direct and strong correlation between greenhouse temperature and H2O mole fraction seen in the closed greenhouse was also observed in this scenario, and even enhanced (correlation = +0.980; Figure 3b). The higher correlation (compared with scenario a) might prove the efficiency of the curtain mechanism in avoiding periods of high temperature stress as with the misfit observed in Figure 1b on March 3, although some stomatal closing still seemed to occur at noon, as indicated by a decrease in H2O accompanied by a slight increase in CO2 mixing ratio. The negative correlation between water vapor and CO2 mole fractions is again obvious during the entire 3-day period, with values similar to those observed in Figure 1b for the closed greenhouse (correlation = -0.843), and considerably higher than those measured in the open greenhouse (Figure 2b).
CO2 mixing ratio gives interesting information on the "closure" of the greenhouse. At night, with the curtains closed, values increase to over 500 mmol mol-1. As the solar radiation increases and photosynthesis begins, when temperatures are still low and the curtains are still closed, the CO2 mixing ratio decreases, reaching values of ~300 mmol mol-1. At noon, the temperature increases and the curtains open, air with more CO2 enters the greenhouse, its mixing ratio increases and, in some cases (see January 15 at 1530 h), even reaches the mean regional value of ~400 mmol mol air-1. In cases where the curtains close again due to colder temperatures, mixing with the outside stops, and photosynthesis causes another decrease in the CO2 mole ratio (see January 15 at 1600–1630 h). Thus, in the closed but automatically controlled greenhouse, diffusion from the "outside" to the "inside" depends strongly on the curtain position.
4.4 Scenario d: open field, peach orchard, August 2022
The last scenario to be discussed is an outdoor orchard during the summer. During the spring and summer of 2022, the instruments were installed in an open peach orchard located in northern Israel, 410 m a.s.l. Figure 4 shows measurements performed in the orchard on 20–22 August 2022.
Unlike the previous scenarios, here the assumptions for the EC calculation of fluxes were valid: the field is large enough and relatively homogeneous. Accordingly, fluxes shown in Figure 4a appear to be significant, and the pattern emerging from the measurements is clear: during the day, CO2 flux is negative (toward the plants) and H2O flux is positive (evaporation from the plants). At night, water vapor flux is close to zero, while CO2 flux exhibits relatively low positive values due to respiration.
The negative correlation between both fluxes was relatively impressive (-0.936). It is interesting to note some specific moments along the graph. For example, for all 3 days at noon, cloudiness formed and there was a reduction in global radiation (Figure 4b). This is particularly salient on August 21, when even a few millimeters of rain were measured in some nearby regions. Immediately after the cloudiness appeared, CO2 flux increased (became less negative) and H2O flux decreased (Figure 4a). When the clouds vanished, the former fluxes were restored. No such clear effect was noticed in scenarios a–c.
As for the temperatures, since it is an open field, we did not expect to see differences between the orchard and the meteorological station nearby (Figure 4b). It is interesting to note that there was some correlation between temperature and water vapor mixing ratio (0.709), but it was considerably lower than that observed in the closed greenhouse. On the other hand, CO2 mixing ratio remained almost constant during the day, with a slight increase at night when the air was stagnant, as noticed decades ago by Keeling (Harris 2010), and mentioned in the introduction.
4.5 Fluxes or concentrations?
As we can see from the different scenarios, when trying to deduce photosynthetic activity, there is no single parameter that can give an accurate indication. In some cases (scenarios a and c, closed or mostly closed greenhouse), the changes in the "total amount" or "concentration" of the gas described by its mixing ratio/molar ratio can yield important information on the gross photosynthesis.
In completely open systems (scenario d), "concentration" does not deliver accurate information due to turbulence and rapid gas exchange, but in those cases, the fluxes (Figure 4a) accurately describe the plant activity. In partially open systems (open greenhouse, scenario b), neither fluxes nor concentrations "work": the fluxes are inaccurate due to the roof that hinders turbulence, and the concentrations fail to describe the process, because the open sides allow relatively rapid gas mixing between the inside and outside.
In looking for parameters that might give an indication of which measurement will be appropriate, we assume that they should be related to turbulence. Turbulence properties are described by the Monin–Obukhov similarity theory (MOST), which provides a set of equations that relate turbulence to atmospheric parameters (Moorhead 2018). A key parameter in MOST is the friction or shear velocity (u*), which relates shear stress to air density, and gives information on the vertical transport of momentum, or turbulence generation (Camuffo 2014). Other relevant parameters associated with eddies in turbulent flow are the turbulence kinetic energy (TKE), defined as the mean kinetic energy per unit mass, the dimensionless stability parameter that is derived from the height of the measurement, and several parameters included in the "Obukhov length" (among them u*, temperature, kinematic heat flux, air density and heat capacity). EddyPro software evaluated all of these parameters based on the EC system described in the materials and methods section.
Figure 5 shows u* and TKE evaluations for scenarios a–d, offering a perspective on the conditions required to decide which parameter to use: when turbulence is hindered by the greenhouse roof and side curtains (scenarios a and c), both u* and TKE are almost zero. On the other hand, when turbulence reigns (scenario d, open field orchard), both values are almost zero during the night, but reach large values during the day. Scenario b has open sides, but also upper coverage. Some turbulence is indeed observed, as reflected by low but noticeable values of both u* and TKE. The measured values during almost the entire 3-day period are 10–20% of the shear velocity and 1–5% of the TKE in the open field. Results for the stability index (not shown in the figure, due to the chaotic fluctuations of the values) also yield similar information: since values of this parameter might be positive or negative depending on the conditions, the regular average value is meaningless. However, the root mean square of the stability index yielded, for scenarios a–d, dimensionless values of 78.9, 1.31, 51.5 and 0.581, respectively. As with the other turbulence-related parameters, lack of turbulence (large values of stability index) are observed for scenarios a and c. The strongest instability (lowest stability index) occurs in scenario d, whereas scenario b exhibits low stability, but still higher than in the open field.