**3. Results and Discussion**

The contribution from this work can be shown in the temperature elements plotted in Figure 7, where the actual temperature pattern for six different heights under agrivoltaic conditions is portrayed, **3. Results and Discussion** 

side.

using 3600 data samples for five consecutive days from 7 a.m. to 7 p.m., daily. Each temperature value came from a thermal sensor (Type K: DS18B20, Maxim Integrated, San Jose, CA, US), starting from Tg, which was the ground surface temperature, up to the bottom of the PV array, (Tb, pv) which was directly glued to the PV array's bottom surface. The other four temperature locations (T1ft,2ft,3ft,4ft) were based on readings from a hanging sensor to measure the surrounding air temperature. portrayed, using 3600 data samples for five consecutive days from 7 a.m. to 7 p.m., daily. Each temperature value came from a thermal sensor (Type K: DS18B20, Maxim Integrated, San Jose, CA, US), starting from Tg, which was the ground surface temperature, up to the bottom of the PV array, (Tb, pv) which was directly glued to the PV array's bottom surface. The other four temperature locations (T1ft,2ft,3ft,4ft) were based on readings from a hanging sensor to measure the surrounding air temperature.

where the actual temperature pattern for six different heights under agrivoltaic conditions is

*Agronomy* **2020**, *10*, x FOR PEER REVIEW 9 of 16

The thermal imager provided some insight into the temperature under agrivoltaic conditions, although the readings might not be too precise because they only showed one spot value at a time. Figure 6 and Video S1 show the temperature values at different locations, i.e., below the PV panel, the surrounding air underneath PV, the surrounding air at plant level, around the leaves and the ground surface temperature taken randomly at different times (5-min intervals). Assumptions were made for the temperature values at each location and level based on the color indicator on the right

**Figure 7.** Temperature trends under agrivoltaic conditions at 1 min intervals (12 h daily). Abbreviations: Tg: Ground temperature; T1ft,2ft,3ft,4ft: Temperature at 1 foot intervals; Tb, pv: PV panel's bottom surface temperature. **Figure 7.** Temperature trends under agrivoltaic conditions at 1 min intervals (12 h daily). Abbreviations: Tg: Ground temperature; T1ft,2ft,3ft,4ft: Temperature at 1 foot intervals; Tb, pv: PV panel's bottom surface temperature.

Based on the temperature values in Table 1, the maximum recorded temperature for T1ft, T2ft and T3ft was 34 °C, 36.5 °C and 36.5 °C, respectively, where, at this height, the plant started growing under agrivoltaic conditions. The value for ∆Tmax was increasing with the plant height–temperature difference (1–2 feet) ranging below 3 °C. The ground temperature (Tg) was considered as the reference value based on its effect on plant seedlings, and Tb (the bottom surface of PV module) as the maximum plant height. Hatfield and Prueger [39] explained that the rate of plant growth and development is heavily dependent on the surrounding temperature (min, max and optimum temperature values) and the annual temperature increment due to global warming over the next 50 years is likely to reach 1.5 °C between 2030 and 2052 [40]. Based on the temperature values in Table 1, the maximum recorded temperature for T1ft, T2ft and T3ft was 34 ◦C, 36.5 ◦C and 36.5 ◦C, respectively, where, at this height, the plant started growing under agrivoltaic conditions. The value for ∆Tmax was increasing with the plant height–temperature difference (1–2 feet) ranging below 3 ◦C. The ground temperature (Tg) was considered as the reference value based on its effect on plant seedlings, and T<sup>b</sup> (the bottom surface of PV module) as the maximum plant height. Hatfield and Prueger [39] explained that the rate of plant growth and development is heavily dependent on the surrounding temperature (min, max and optimum temperature values) and the annual temperature increment due to global warming over the next 50 years is likely to reach 1.5 ◦C between 2030 and 2052 [40].


**Table 1.** Values of temperature difference, ∆T, (in °C) based on a 1 foot height distribution. **Table 1.** Values of temperature difference, ∆T, (in ◦C) based on a 1 foot height distribution.

Abbreviations: Tg: Ground temperature; T1ft,2ft,3ft,4ft: Temperature at 1 foot intervals; Tb, pv: PV panel's bottom surface temperature.

Based on Equation (1) and an online calculator software, the values for VPD are summarized in Table 2. The value for T1ft was used to represent the designated surrounding air temperature (Ta) because the location was at par with the plant at a 1 foot height and touching the polybags and soil.


**Table 2.** Vapor pressure density (VPD) calculations based on 1 foot height under agrivoltaic conditions.

Abbreviations: Ta: Ambient temperature; % RH: Relative humidity; SVP: Saturated vapor pressure; VP: Vapor pressure; VPD: Vapor pressure density.

The optimum value for VPD under a greenhouse condition ranges from 0.45 kPa to 1.25 kPa, ideally sitting at around 0.85 kPa [31]. For agrivoltaic conditions, the VPD value ranged between 2.005 kPa (max) to 0.548 kPa (min), with an average value of 1.072 kPa.

For the temperature analysis, the field data measured were segregated into five sampling hours (daily) with different temperature levels, as shown in Table 3.


**Table 3.** Analysis of temperature distributions based on sampling hours.

Abbreviations: Tg: Ground temperature; T1ft,2ft,3ft,4ft: Temperature at 1 foot intervals; Tb, pv: PV panel's bottom surface temperature.

Based on Table 3 and R programming, the heat stress contour throughout the five sampling hours was plotted as shown in Figure 8.

An illustration of heat stress occurrences in % value with respect to the 1 foot height–temperature level under agrivoltaic conditions is shown in Figure 8. These field data were further analyzed as shown in Figure 9, where dependencies on the bottom of the PV panel and at a 4 foot height can be observed.

*Agronomy* **2020**, *10*, x FOR PEER REVIEW 11 of 16

**Figure 8.** Heat stress occurrences (%) at five sampling hours. **Figure 8.** Heat stress occurrences (%) at five sampling hours. analyzed as shown in Figure 9, where dependencies on the bottom of the PV panel and at a 4 foot height can be observed.

**Figure 9.** Field observation for heat stress directly underneath the PV arrays. Abbreviations: Ta: Ambient temperature; Tg: Ground temperature; T1ft,2ft,3ft,4ft: Temperature at 1 foot intervals; Tb, pv: PV panel's bottom surface temperature. **Figure 9.** Field observation for heat stress directly underneath the PV arrays. Abbreviations: Ta: Ambient temperature; Tg: Ground temperature; T1ft,2ft,3ft,4ft: Temperature at 1 foot intervals; Tb, pv: PV panel's bottom surface temperature.

Based on Figure 8, the percentage of heat stress occurrences shows at what specific time in the day the plant will possibly experience a high surrounding temperature, above the normal ambient temperature. Based on the data sample, the highest heat stress occurred at a 4 foot height during peak sun and moderate sun (afternoon), with more than 23% heat stress points, as shown in Table 4. This is due to the bottom of the PV panel producing a much higher temperature after the photonic conversion and heat dissipation process. The ground heat's effect in this agrivoltaic condition was Based on Figure 8, the percentage of heat stress occurrences shows at what specific time in the day the plant will possibly experience a high surrounding temperature, above the normal ambient temperature. Based on the data sample, the highest heat stress occurred at a 4 foot height during peak sun and moderate sun (afternoon), with more than 23% heat stress points, as shown in Table 4. This is due to the bottom of the PV panel producing a much higher temperature after the photonic conversion and heat dissipation process. The ground heat's effect in this agrivoltaic condition was relatively low due to the PV array shading, as per temperature values for T<sup>g</sup> until T2ft, thus, it can be assumed that no heat stress was caused by this.


A two-sample proportion test and a Chi-square test were used as the statistical approaches as shown in Tables 5 and 6, respectively.

**Table 5.** Count of heat stress (Th) cases across temperature–height levels during peak sun.


**Table 6.** Chi-square test for difference in proportions of heat stress (Th) occurrence during peak sun.


Based on the Chi-square test, T4ft had a higher percentage of heat stress occurrence than Tb,pv during peak sun at 99% confidence level (*p* < 0.00001). The same test was conducted for height level during moderate sun (afternoon) and these results also proved that T4ft had a higher percentage of heat stress occurrence than Tb,pv during moderate sun (afternoon) at 99% confidence level (*p* < 0.00001).

Based on the correlations of Tb,pv and T4ft towards heat stress (Th) under agrivoltaic conditions, a summary of the findings of both the minimum and maximum values of heat stress, Th,min and Th,max, is modelled as shown in Table 7. Some preliminary assessments were conducted to assess the fitness of data for regression modelling and the findings are displayed in Figures S1a–d, S2a–d, S3a–b, S4a–b and Table S7. Since all assumptions were fulfilled, regression models were developed and detailed findings are presented in Table S1–S6 which were simplified into Tables 7 and 8. The coefficient of determination (R squared) was 0.739, which indicates that 73.9% of the variation in Th,min and Th,max could be explained by the variation in both Tb, pv and T4ft, and both the Th,min and Th,max models were significantly fit at a 99% confidence level (*F* = 4724.462, *p*-value < 0.001).

**Table 7.** Regression statistics and analysis of variance (ANOVA).


Multiple R = 0.860; R Square = 0.739; Adjusted R Square = 0.739; Standard Error = 1.205; Observation Counts = 3335.


**Table 8.** Individual t-test on independent variables.

A t-test on independent variables, as shown in Table 8, confirmed that both Tb,pv and T4ft significantly affected the Th,min and Th,max at 99% confidence level (*tTb, pv* = −57.141, *tT4ft* = 78.155; *p*-value < 0.001). Hence, both were significant predictors of Th,min and Th,max. Meanwhile, a unit increase of Tb, pv, Th,min and Th,max would decrease by 0.293 ◦C, and a unit increase in T4ft would increase Th,min and Th,max by 0.987 ◦C.

Th,min and Th,max could be expressed by the following new equations:

$$\rm{T}\_{\rm{h,min}} = 16.553 - 0.293T\_{\rm{b,pv}} + 0.987T\_{\rm{4ft}} \tag{3}$$

$$\rm{T}\_{h,max} = 21.553 - 0.293T\_{b,pv} + 0.987T\_{4\text{ft}} \tag{4}$$

Or both equations could be simplified into a heat stress temperature model:

$$\text{T}\_{\text{h}} \text{ (Heat stress temperature)} = [16.553, 21.553] - 0.293 \text{T}\_{\text{b,pv}} + 0.987 \text{T}\_{\text{4ft}} \tag{5}$$

### **4. Conclusions**

As a major source of renewable energy, many photovoltaic farms have now been constructed in the world. The agrivoltaic system is a further concept that aims to combine commercial agriculture and photovoltaic electricity generation in the same space, in order to maximize crop production while addressing land management and sustainability issues.

This paper has presented the field measured data of ambient temperature profile and the heat stress occurring directly underneath solar photovoltaic (PV) arrays (monocrystalline-based) in a tropical climate condition (in Malaysia). With reference to the plant heat stress at 10 ◦C to 15 ◦C above the ambient temperature, the percentage of heat stress occurrences was the highest at a 4 foot height during peak sun and moderate sun (afternoon), with more than 23% heat stress points. It has also been found that the ground heat effect in this agrivoltaic condition was relatively low due to the PV array shading. A heat stress model for ground-mounted agrivoltaic conditions has been developed. It has been found that the coefficient of determination (R squared) for the model is 0.739, indicating that 73.9% of variation in Th,min and Th,max could be explained by the variations in both Tb, pv and T4ft. Both Th,min and Th,max models were significantly fit at 99% confidence level. This paper has contributed to the understanding of plant physiological processes in response to environmental conversion factors. The model developed could also be used for further exploring the integration of crop cultivation and PV energy generation for optimum land use.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4395/10/10/1472/s1, Table S1: Regression statistics of Th,min Model, Table S2: Analysis of variance (ANOVA) Th,min Model, Table S3: Individual t-test on independent variable Th,min Model, Table S4: Regression statistics Th,max Model, Table S5: Analysis of variance (ANOVA) Th,max Model, Table S6: Individual t-test on independent variable Th,max Model, Table S7: Variance inflation factor (VIF) for all independent variables, Figure S1: Boxplots for outlier detection. (a) Tb,pv; (b) T4ft; (c) Th,min; (d) Th,max, Figure S2: Scatter plots between dependent and independent variables for linearity. (a) Tb,pv against Th,min; (b) T4ft against Th,min; (c) Tb,pv against Th,max; (d) T4ft against Th,max, Figure S3: Normal QQ plot for residuals for normality; (a) Th,min model; (b) Th,max model, Figure S4: Residuals against fitted values plots for homoscedasticity; (a) Th,min model; (b) Th,max model.

**Author Contributions:** Conceptualization, N.F.O., M.E.Y., and A.S.M.S.; Methodology, N.F.O.; Software, A.H.J., N.F.O.; Validation, formal analysis, and investigation, N.F.O., M.E.Y., A.S.M.S., and A.H.J.; Resources, M.E.Y. and A.S.M.S.; Data curation, N.F.O. and A.H.J.; Writing—original draft preparation, N.F.O. and M.E.Y.; Writing—review and editing, A.S.M.S., J.N.J., H.H., M.F.S., G.C., and A.J.; Visualization, N.F.O.; Supervision, M.E.Y. and A.S.M.S.; Project administration, N.F.O.; Funding acquisition, M.E.Y. and A.S.M.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Ministry of Energy, Science, Technology, Environment and Climate Change (MESTECC) under the MESITA (Malaysia Energy Supply Industry Trust Account) Research Fund (Vote no. 6300921), and the Research Management Center (RMC), University Putra Malaysia, for the approval of research funding under the IPS Putra Grants Scheme (Vote no. 9667400).

**Acknowledgments:** The authors delegate our thanks to the Ministry of Energy, Science, Technology, Environment and Climate Change (MESTECC) under the MESITA (Malaysia Energy Supply Industry Trust Account) Research Fund (Vote no. 6300921) and the Research Management Center (RMC), University Putra Malaysia, for the approval of research funding under the IPS Putra Grants Scheme (Vote no. 9667400).

**Conflicts of Interest:** The authors declare no conflict of interest.

## **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Agronomy* Editorial Office E-mail: agronomy@mdpi.com www.mdpi.com/journal/agronomy

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com ISBN 978-3-0365-3289-9