Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/issn/15375110

Research Paper

Spatial and temporal variation of photosynthetic


photon flux density within agrivoltaic system in
hot arid region of India

Priyabrata Santra*, Hari Mohan Meena, O.P. Yadav


ICAR-Central Arid Zone Research Institute, Jodhpur, Rajasthan, 342003, India

article info
Ground-mounted photovoltaic (PV) arrays of an agrivoltaic system (AVS) creates shade in
Article history: the interspace areas as well as below PV areas of the AVS. Therefore, availability of
Received 22 October 2020 photosynthetic photon flux density (PPFD) and daily light integral (DLI) in the AVS is
Received in revised form negatively affected, which restricts optimum crop growth and yield. Therefore, in this
7 June 2021 study, spatial and temporal variation of PPFD and DLI in the AVS (105 kWp) developed at
Accepted 20 June 2021 Jodhpur, India were quantified. Among three designs of PV arrays in the AVS, the pro-
Published online 7 July 2021 portion of shaded area in interspace was found highest in single row design and lowest in
triple row design. The amount of PPFD varied widely in the AVS and even within the
Keywords: shaded area. Following semivariogram and kriging approach, spatial variation of PPFD and
Agrivoltaic system (AVS) DLI was quantified within the AVS. Highest amount of PPFD was observed near noon in
Photosynthetic photon flux density unshaded areas (1295 mmol m2 s1) whereas lowest amount was observed during morning
(PPFD) at below PV array (24 mmol m2 s1). Based on the maps of PPFD and DLI, three distinct
Photosynthetically active radiation zones in the AVS were identified: (i) below PV array (DLI ¼ 0e5 mol m2 d1), (ii) interspace
(PAR) area with partial shade (DLI ¼ 5e20 mol m2 d1) and (iii) interspace areas with no shade
Daily light integral (DLI) (DLI ¼ 20e30 mol m2 d1). Knowledge of spatial and temporal variation of PPFD and DLI
Spatial variation may help in selection of suitable crops for the AVS.
Shade of PV structure © 2021 IAgrE. Published by Elsevier Ltd. All rights reserved.

of AVS was first proposed by Goetzberger and Zastrow (1982),


1. Introduction and was later successfully studied experimentally by Dupraz
et al. (2011) and Marrou, Guilioni, et al. (2013). During the
An agrivoltaic system (AVS) is defined as combined produc- last 3e4 years, there has been growing interest in utilising the
tion of photovoltaic (PV) electricity and food from a single land potential of AVS in different parts of the world (Adeh et al.,
unit. In the AVS system, the interspaces between ground- 2019; Amaducci et al., 2018; Barron-Gafford et al., 2019;
mounted PV arrays are used for crop cultivation. Even the Dayananda, 2018; Dinesh & Pearce, 2016; Dos Santos, 2020;
areas below PV modules can also be used for the same purpose Elamri et al., 2018; Irie et al., 2019; Kim et al., 2020; Loots, 2018;
if the ground-mounted PV array structure is high enough to Majumdar & Pasqualetti, 2018; Malu et al., 2017; Marrou, 2019;
perform agricultural activities below it with ease. The concept Othman et al., 2018; Ravi et al., 2016; Santra et al., 2017, 2018a;

* Corresponding author.
E-mail address: Priyabrata.Santra@icar.gov.in (P. Santra).
https://doi.org/10.1016/j.biosystemseng.2021.06.017
1537-5110/© 2021 IAgrE. Published by Elsevier Ltd. All rights reserved.
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 75

Nomenclature TC Time correction factor


xi Coordinates of location along x-axis, m
a Semivariogram parameter, range, describing
yi Coordinates of location along y-axis, m
spatial variation of PPFD, m
zðsi Þ Measured value of PPFD at location si, mmol
AVS Agrivoltaic system
m2 s1
C Semivariogram parameter, partial sill, describing
zðsi þhÞ Value of PPFD at lag distance h from the location si,
spatial variation of PPFD, [mmol m2 s1]2
mmol m2 s1
C0 Semivariogram parameter, nugget, describing
zobs Mean of observed values of PPFD, mmol m2 s1
spatial variation of PPFD, [mmol m2 s1]2
zpred Mean of predicted values of PPFD, mmol m2 s1
d Julian day, day
b
z ðsi Þ Predicted values of PPFD at location si, mmol m2
D0 Maximum shade length, m
s1
Dx Shade length in x-axis, m
b
z ðs0 Þ Kriging estimate of PPFD at unmeasured location
Dy Shade length in y-axis, m
s0, mmol m2s1
DLI Daily light integral, mol m2 d1
var Variance
EoT Empirical equation of time
d Solar declination, degree
h Lag distance, m
DGMT Time difference with Greenwich Mean Time, h
ht Height of PV array from ground surface, m
gðhÞ Semivariogram value of PPFD at lag distance h,
LCCC Lin's concordance correlation coefficient
[mmol m2s1]2
LST Local solar time, hour
li Kriging weight for the data observations z (si)
LSTM Local standard time meridian, degree
located at si
LT Local time, hour
4s Solar azimuth angle, degree
n Number of sampling locations
j Latitude of location, degree
N(h) number of pairs of data locations within a given
r Pearson correlation coefficient
class of lag distance h
s2E ðs0 Þ Error variance of kriged prediction of PPFD at
PAR Photosynthetically active radiation, mmol m2 s1
location s0, [mmol m2 s 1]2
PPFD Photosynthetic photon flux density, mmol m2 s1
sobs Standard deviation of observed values of PPFD,
PV Photovoltaic
mmol m2 s1
RMSE Root mean squared error
spred Standard deviation of predicted values of PPFD,
SEF Solar energy farming
mmol m2 s1
si Coordinates (xi ; yi ) of measurement location, (m,
qs Solar zenith angle, degree
m)
u hour angle at local solar time, degree
s0 Coordinates (x0 ;y0 ) of unmeasured location, (m, m)

Weselek et al., 2019). Solar PV systems have been installed in PV greenhouses and therefore should not be used
agricultural lands with the main focus of growing crops in interchangeably.
association with solar panels; however, nomenclature of the Arid regions or drylands of the world receiving enough
system other than AVS has also been reported in the literature solar irradiation are ideal place for establishment of AVS
e.g. agrophotovoltaic system (Schindele et al., 2020; Weselek (Barron-Gafford et al., 2019), though high wind regime in the
et al., 2019), agri-voltaic system (Santra et al., 2017), solar en- region restricts the mounting height of PV array above ground
ergy farming (SEF) (Al-Saidi & Lahham, 2019), co-location of level, otherwise it may be dislodged by high wind action.
solar and agriculture (Ravi et al., 2016) etc. Combining PV Moreover, cleaning of deposited dust from the PV module is
generation and agricultural production together in AVS has difficult in case of high mounting PV arrays of the AVS since
led to improvement of land productivity by about 40e70% erosion of topsoil by wind action is a regular phenomenon in
(Dupraz et al., 2011; Santra, Singh, Jain, & Yadav, 2018; Valle arid regions. Shade created by opaque PV modules of ground
et al., 2017). Further, AVS system has also been found to be mounted AVS may have significant impact on plant growth
environmentally friendly and therefore it may be thought as a and yield (Kadowaki et al., 2012; Poorter et al., 2019) and this
suitable option for mitigating climate change effects specif- effect will be more pronounced in the case of low mounting PV
ically in drylands (Santra et al., 2018; Leon & Ishihara, 2018; array than high mounting structure. The occurrence of shade
Barron-Gafford et al., 2019). PV greenhouses are other aspects is intermittent in nature since it depends on solar zenith and
of combining PV and agriculture where controlled cultivation azimuth angle, but the cumulated duration of shade at
of horticultural crops is carried out inside protected structures different locations in AVS is an important factor to govern
and PV modules on the roof of the structure are either partially crop growth and yield in AVS. Shade does not always nega-
or fully used to meet the energy requirement for controlled tively affect crop growth if the selected crop can tolerate a
cultivation (Li et al., 2017; Marucci & Cappuccini, 2016; Marucci certain degree of shade (Sekiyama, 2019; Sekiyama &
et al., 2018; Wang, Wu, et al., 2017; Yano & Cossu, 2019; Yano Nagashima, 2019). Sometimes, the shade may be beneficial
et al., 2010). However, the AVS system is quite different from to crops, especially during hot summer days, by limiting
76 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

exposure of crops to excessive light, by improving the soil 2017a, 2018; Wang & Sun, 2018). Since, low mounting struc-
water regime for plant growth and by modifying microclimate ture is generally followed in arid environment or dryland, the
(Adeh et al., 2018; Marrou et al., 2013b, 2013c; Oleskewicz, knowledge on pattern of shade and availability of PPFD and
2020; Patel & Chauhan, 2018). DLI within AVS is important.
Therefore, availability of photosynthetic photon flux den- The present study was aimed to quantify the spatial and
sity (PPFD) and its spatial and temporal variation plays an temporal variability of PPFD and DLI in the 105 kWp AVS
important role in successful crop production in AVS. PPFD is established at Jodhpur, India. The specific objectives were to
defined as the number of photosynthetically active photons quantify the availability of PPFD in the interspace and below
that fall on a given surface each second and includes photons PV areas of the AVS during the winter solstice period and to
in 400e700 nm region of the solar energy spectrum, which develop spatial maps of shade coverage, PPFD and DLI in the
constitute about 42e49% of the total energy of the whole solar AVS using geostatistical techniques.
spectra (Faust & Logan, 2018; Moretti & Marucci, 2019; Zhu
et al., 2008). The photosynthesis process utilises only 3e4%
of the total incident solar energy although this varies across 2. Materials and methods
plant types (Zhu et al., 2008). Therefore, reduced availability of
PPFD under shade negatively affects photosynthesis and plant 2.1. Design of AVS
growth (Gao et al., 2010; Kla € ring & Krumbein, 2013). Often,
PPFD is expressed as photosynthetically active radiation (PAR) The AVS of 105 kWp was installed at the experimental solar
and its cumulated availability is known as daily light integral yard of ICAR-Central Arid Zone Research Institute (ICAR-
(DLI) (Faust et al., 2005; Hodges et al., 2016; Suthaparan et al., CAZRI), Jodhpur. The geographical location of the study site
2017). The situation of the shade created by the PV structure was 26 150 27.8200 N and 72 590 34.5700 E. The AVS was installed
in AVS can be compared to the shade created by trees in at three blocks with an area of 32  32 m for each. The PV
agroforestry systems, where performance of crop growth de- system in each block was 35 kWp and the length of the PV
pends on the shade of associated trees in the system (Bellow & array in each block was 30.75 m. Power output and dimension
Nair, 2003; Meloni & Sinoquet, 1997; Sivakumar & Virmani, of a single PV module was 260 Wp and 1.64 m  0.992 m,
1984). Therefore, knowledge of spatial and temporal varia- respectively. Design parameters of PV module installations in
tion of PPFD in AVS may help to select suitable crops for the each block are given in Fig. 1. Inclination of the PV mounting
AVS. Spatial variation of available PPFD inside a greenhouse structure from ground surface was 26 , which was equal to the
structure has been quantified by Guertal and Elkins (1996) by latitude of the experimental site. At the low end of the PV
applying a geostatistical approach. Later, several attempts mounting structure, ground clearance was 0.5 m. Interspace
have been made to quantify spatial and temporal variation of distance between two PV arrays in block1, block 2 and block 3
PPFD inside PV greenhouse structure (Castellano et al., 2016; were 2.95 m, 5.55 m and 8.90 m, respectively.
Cossu et al., 2017; Moretti & Marucci, 2019). However, studies The design of the PV array was different in different blocks
to quantify spatial and temporal variation of PPFD in AVS and of the AVS: (i) single row PV array in block 1 (ii) double row PV
for low mounting PV arrays have been limited (Wang et al., array in block 2 and (iii) triple row PV array in block 3. The

Fig. 1 e Design of ground mounted PV array structure in agrivoltaic system (AVS) at ICAR-CAZRI, Jodhpur.
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 77

number of PV arrays in block 1, block 2 and block 3 were 6, 3 2.3. Shade calculation in AVS
and 2, respectively to keep the total size of the PV systems in
each block as 35 kWp. In block 1, PV modules were installed Shade of a ground-mounted solar PV array depends on the
with full density in three PV arrays and half density in the sun's position, which is mainly defined by solar azimuth angle
other three PV arrays. In block 2 and 3, PV density was less and zenith angle (Duffie & Beckman, 2013). Graphical repre-
than its full capacity. The double row model in block 2 con- sentations of solar azimuth and zenith angle and the shade
sisted of the bottom row with full density of PV modules and created by the ground-mounted PV structure are illustrated in
the top row with 60% density of PV module. The triple row Fig. 3.
model in block 3 consisted of two bottom rows with full Solar zenith angle (qs) was calculated using:
density of PV modules and the top row with 60% density of PV
modules. Installation designs of full density in single row PV cosðqs Þ ¼ sinðjÞsinðdÞ þ cosðjÞcosðuÞ (1)
array of block 1, fullþ60% density in double row PV array of
block 2, and fullþfullþ60% density in triple row PV array of where, j is latitude of location, d is the solar declination, u is
block 3 are depicted in Fig. 2. Designs of the other two PV ar- the hour angle at local solar time.
rays in the AVS are not presented here since no observations Solar azimuth angle (4s) was calculated using:
of PPFD were recorded under these PV arrays. Overall, there
sinðdÞ:cosðjÞ  cosðdÞ:sinðjÞ:cosðuÞ
were 28 PV modules in of the full density rows and 17 PV cosð4s Þ ¼ (2)
cosðqs Þ
modules in the 60% PV density rows.
Solar declination angle (d) was calculated using:
2.2. Solar irradiance data of Jodhpur
d ¼ 23:45 sinðBÞ (3)

Solar irradiance (W m2) at the experimental site was


360
measured at 5-min interval during 2016e2018 using an in- B¼ ðd  81Þ (4)
365
built pyranometer in automatic weather station which was
installed near the AVS. Average hourly variation of solar en- where, B is a factor which corrects the eccentricity of earth's
ergy (Wh m2) was calculated from the raw data of solar orbit and earth's axial tilt, d is the Julian day in a year with
irradiance (W m2) for each month in a year. The radiometric values from 1 to 365 at 1st January and 31st December,
data on solar energy was further converted to photosynthetic respectively.
photon flux density (PPFD) using a conversion factor of Hour angle at local solar time was calculated using:
0.007265 mol (400e700 nm) Wh1 (400e2700 nm) considering
that 45% of solar energy spectrum is available in photosyn- u ¼ 15ðLST  12Þ (5)
thetically active radiation (PAR) (400e700 nm) region and the
conversion factor from radiometric unit to quantum unit is TC
LST ¼ LT þ (6)
4.48 mmol J1 (Faust & Logan, 2018). 60

Fig. 2 e Agrivoltaic system (AVS) at ICAR-CAZRI, Jodhpur; (aec) design of PV array in three field blocks of the AVS and (def)
field photographs of the installed AVS in block 1, block 2 and block 3, respectively.
78 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

Fig. 3 e Schematic diagrams of solar azimuth and zenith angle and shade created by ground mounted PV structure (In the
top figure, solid lines are on XeY plane or ground surface whereas dotted lines are above XeY plane). (For interpretation of
the references to color in this figure legend, the reader is referred to the Web version of this article.)

TC ¼ 4ðLongitude  LSTMÞ þ EoT (7) Dy ¼ D0 cosð180  4s Þ (12)

where, Dx is shade length in x-axis and Dy is shade length in y-


LSTM ¼ 15DGMT (8)
axis of x-y plane.

EoT ¼ 9:87 sinð2BÞ  7:53 cosðBÞ  1:5 sinðBÞ (9) 2.4. Measurement of PPFD in AVS
where, LST is the local solar time, LT is the local time, TC is the
time correction factor in minutes which accounts for the The PPFD at multiple locations in each selected PV array design
variation of LST within a time zone, LSTM is the local standard was measured using a line quantum sensor (Apogee, MQ-301,
time meridian, DGMT is the time difference with Greenwich Series#1178). The line quantum sensor consisted of 10 quan-
Mean Time in hour (5.5 h in India), EoT is the empirical tum sensors spaced 5 cm apart. Each quantum sensor mea-
equation of time in minutes that corrects for the eccentricity sures the quantity of photons available in the sunlight in the
of the earth's orbit and earth's axial tilt. spectral range of 410e655 nm. In-built wavelength dependent
Using the solar zenith angle, shade of ground mounted PV adjustment coefficients of 0.5 or above were used to quantify
structure was calculated as follows: the total number of photons between 410 and 655 nm. Blue light
has lower adjustment coefficients compared to red light.
ht Spatial averaging of 10 sensors was recorded as a single output
D0 ¼ (10)
tanð90  qs Þ using a hand-held meter attached to the instrument. During
where, D0 is the maximum shade length for a particular solar measurements of PPFD in field at each location, the instrument
zenith angle, qs and ht is the height of PV array from ground was held 10 cm above the ground surface in a west-east di-
surface. The spatial coverage of shade in x-y plane on the ground rection. Geographical coordinates of each measurement loca-
surface was further calculated by converting maximum shade tions were recorded (Fig. 4). The numbers of sampling locations
length to perpendicular shade length in x-y plane as follows: in single row PV array, double row PV array and triple row PV
array were 66, 142 and 209, respectively. PPFD was measured
Dx ¼ D0 sinð180  4s Þ (11) during cloud-free day around winter solstice in northern
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 79

Fig. 4 e Measurement locations of photosynthetic photon flux density (PPFD) in the agrivoltaic system (AVS) at ICAR-CAZRI,
Jodhpur (dotted line is the border line between below PV area and interspace area of AVS).

hemisphere e.g., on 20th December 2018 in double row PV array 2.5. Mapping spatial variation of PPFD in AVS
(block 2), on 22nd December 2018 in triple row PV array (block 3)
and on 24th December 2018 in single row PV array (block 1). Spatial variation of PPFD in the AVS was determined by
Measurements were recorded around winter solstice because calculating the semivariogram followed by kriging. Semi-
day length was minimum during this period in northern variogram, gðhÞ, measures the average dissimilarity in a
hemisphere and thus the magnitude of shade was greatest and property between data locations separated by a vector h
its effect on availability of PPFD was also greatest. PPFD was (Webster & Oliver, 2007).
recorded during three periods on each measurement day: (i)
1 X
NðhÞ
9:00e10:00 (morning), (ii) 12:00e13:00 (noon), and (iii) 2
gðhÞ ¼ ½zðsi Þ  zðsi þ hÞ (13)
15:00e16:00 (afternoon). 2NðhÞ i¼1
80 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

where, N (h) is the number of pairs of data locations within a 2.6. Daily light integral map in AVS
given class of lag distance, z (si) is the value of the variable at
the location si with coordinates of (xi, yi), z (si þ h) is the value The PPFD values indicated the instantaneous availability of
of the variable at a lag of h from the location si. photosynthetically active radiation, which varied widely in a
Kriging estimates of PPFD at unmeasured locations, zbðs0 Þ, day. Therefore, an integrated parameter, daily light integral
were obtained using weighted linear combinations of known (DLI), was used to quantify the total availability of light in a
PPFD values, z (si), located within a neighbourhood centred day and was expressed as mol m2 d1. DLI was quantified by
around s0 (Webster & Oliver, 2007). summing up the cumulated PPFD for three measured sub
periods in a day (morning, noon, and afternoon). Using the
X
nðs 0Þ
above procedure, maps of PPFD for 9:00e10:00 (morning),
b
z ðs0 Þ ¼ li zðsi Þ (14)
i¼1 12:00e13:00 (noon) and 15:00e16:00 (afternoon) were con-
verted to a DLI map.
where li is the weight assigned to datum zðsi Þ located within a
given neighbourhood centred on s0. Weights for n number of
neighbourhood points were chosen as such to minimise the
3. Results and discussion
estimation or error variance, s2E ðs0 Þ ¼ Varf zbðs0 Þ  zðs0 Þgunder
the constraint of no-bias of the estimator. Semivariogram
3.1. Solar irradiance and estimated PPFD at Jodhpur
calculation and kriging were done using ‘gstat’ package of R
(Pebesma, 2003; R Core Team, 2019).
Average monthly variation of solar irradiation at Jodhpur
Accuracy of spatially predicted PPFD was checked using k-
during 2016e2018 is presented in Fig. 5. The highest solar
fold cross validation (k ¼ 10). In this approach, the whole
irradiance was observed during May (7.3 kWh m2 d1) and the
dataset was divided into k-folds and then (k-1) fold datasets
lowest in December (3.7 kWh m2 day1). The duration of
were used as training dataset and the kriging system based on
receiving significant amount of solar irradiation (>100 Wh
the training data was tested on the kth fold as validation
m2) was about 11 h (from 7:00 to 18:00) during summer
dataset. The procedure was repeated till each of k folds got a
months (April, May and June) and about 9 h (from 8:00 to 17:00)
chance to become validation dataset once. The following
during winter months (November, December and January).
validation statistics were calculated to judge the accuracy of
Annual average solar irradiance at the experimental site was
PPFD maps.
5.2 kWh m2.
   Monthly and daily temporal variation of available PPFD at
P
n
zðsi Þ  zobs b z ðsi Þ  zpred Jodhpur is presented in Table 1. Wide variation in availability of
i¼1
r ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2 (15) PPFD was observed across different months and across
P n Pn
zðsi Þ  zobs b
z ðsi Þ  zpred different times in a day. Peak of daily variation in PPFD was
i¼1 i¼1
observed for 12:00e13:00 in each month and the peak value
varied from 1206 mmol m2 s1 during December to
2rsobs spred
LCCC ¼  2 (16) 1970 mmol m2 s1 during May. Monthly variation of DLI
zobs  zpred þ s2obs þ s2pred showed a range from 27 mol m2 d1 during December to
53 mol m2 d1 during May. Poorter et al. (2019) reported the
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi monthly averaged DLI value in the range of 20e40 mmol m2 s1
1X n
2 in tropics, 15e60 mmol m2 s1 at 30 latitude and
RMSE ¼ ½zðsi Þ  b z ðsi Þ (17)
n i¼1

1X n
bias ¼ ½zðsi Þ  b
z ðsi Þ (18)
n i¼1

b i Þis the predicted values with variance s2 at the


where, Zðs
location si, and n is the number of sampling location. The
Pearson correlation coefficient (r) measured the precision of
the relationship. Lin's concordance correlation coefficient
(LCCC) (Lin, 1989), on the other hand, was a single statistic
that evaluated both the accuracy and precision of the rela-
tionship. A LCCC value equal to 1 would indicate perfect
positive agreement between observed and predicted values
whereas 1 indicated perfect negative agreement. The RMSE
statistics estimated the accuracy of prediction (e.g., larger
RMSE values indicated less accuracy of prediction). Bias
indicated the mean error of prediction and a value of zero
indicated unbiased prediction. Cross-validation was per-
formed using ‘gstat’ package of R (Pebesma, 2003; R Core Fig. 5 e Average hourly variation of solar irradiation at
Team, 2019). Jodhpur, India for different months (Period: 2016e2018).
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 81

Table 1 e Temporal variation of photosynthetic photon flux density (PPFD) (mmol m¡2 s¡1) and daily light integral (DLI) (mol
m¡2 d¡1) for different months at Jodhpur, India.
Time PPFD (mmol m2 s1)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
5:00e6:00 0 0 0 1 6 8 6 1 0 0 0 0
6:00e7:00 0 1 12 62 138 136 91 48 34 12 3 0
7:00e8:00 30 76 201 340 494 454 312 249 267 205 124 52
8:00e9:00 261 391 537 709 892 823 611 550 616 573 456 328
9:00e10:00 582 770 896 1036 1306 1170 903 870 940 926 809 653
10:00e11:00 857 1118 1267 1348 1588 1502 1128 1148 1236 1177 1104 949
11:00e12:00 1098 1357 1506 1549 1864 1710 1285 1326 1369 1388 1266 1146
12:00e13:00 1248 1450 1582 1628 1970 1821 1434 1326 1531 1460 1329 1206
13:00e14:00 1214 1406 1526 1562 1911 1776 1420 1302 1457 1374 1224 1124
14:00e15:00 1036 1266 1345 1373 1695 1607 1268 1188 1266 1144 999 949
15:00e16:00 800 954 1055 994 1382 1327 1061 1023 937 765 698 670
16:00e17:00 437 611 689 636 936 944 767 691 599 409 340 333
17:00e18:00 118 250 307 340 411 415 438 386 295 108 50 49
18:00e19:00 3 17 43 66 139 153 165 110 38 3 0 0
19:00e20:00 0 0 0 1 10 18 14 6 0 0 0 0
DLI (mol m2 d1) 28 35 39 42 53 50 39 37 38 34 30 27

1e40 mmol m2 s1 at 60 latitude. Thus, the DLI value for during most of the period in a year (Table 1) and thus meeting
different months at Jodhpur (latitude ¼ 26 ) as presented in the light requirement of field crop was not a constraint in the
Table 1 coincides well with the reported values in literature. region. Rather, excess exposure to light, which is quite common
PPFD is a critical variable governing the rate of photosyn- in arid environments e.g., at Jodhpur, negatively affects crop
thesis in plants. It has been reported that an increase in PPFD growth and yield due to heat stress and associated changes in
results in an increase in photosynthesis rate, which is further microclimate. Therefore, to cut off exposure to light that ex-
reflected in crop growth and yield (Baligar et al., 2005; Chauhan ceeds the optimum requirement may be a strategy to achieve
et al., 2013; Hidaka et al., 2013; Kla € ring & Krumbein, 2013; optimum crop growth and yield in arid environments. How-
Ohashi-Kaneko et al., 2007; Sivakumar & Virmani, 1984). ever, plants should not be denied receiving the minimum
However, the increasing trend in photosynthesis has generally amount of PPFD to overcome the light compensation point,
been observed up to a critical level of PPFD, which is known as above which net photosynthesis rate becomes positive. The
the light saturation point, beyond which the increasing trend information on PPFD requirement at light compensation point
was flattened. The PPFD corresponding to the light saturation is limited in the literature and specifically for field crops. Shade
point has also been reported to vary between different crop on the ground surface created by an opaque object (PV struc-
types (Yano & Cossu, 2019) (Table 2). However, most of the ture) drastically reduces the availability of PPFD and thus
literature has reported the required PPFD for horticultural crops negatively affects plant growth whereas availability of PPFD
and such information is limited for field crops. It is because of above light saturation point does not positively influence plant
the difficulty to conduct controlled experiments on quantifying growth. Therefore, optimum availability of PPFD in between
the light requirement of field crops. Experimental efforts on light compensation point and light saturation point is always
quantifying the light requirement of horticultural crops were beneficial to crop growth and yield and hence its availability in
feasible because a controlled experimental set up is compara- AVS needs to be ascertained to achieve optimum crop yield.
tively easy to build inside greenhouses using artificial lighting.
Overall, it was observed that PPFD at light saturation point 3.2. Spatial and temporal variation of shade in AVS
ranged from 450 to 1630 mmol m2 s1 with an average value of
992 mmol m2 s1 for horticultural crops. This amount of PPFD Spatial and temporal variation of shade created by ground-
was abundantly available at the experimental site at Jodhpur mounted PV arrays of the AVS during winter solstice is

Table 2 e List of crops with required amount of PPFD for optimum growth and yield.
Crops PPFD (mmol m2 s1) requirement Reference
Minimum Maximum
Cucumber (Cucumber sativa L.) 250 450 Moretti and Marucci (2019)
Lettuce (Lactuca sativa L.) e 450 Sun et al. (2016)
Tomato (Lycopersicon esculentum) 450 750 Moretti and Marucci (2019)
Muskmelon e 1630 Chen et al. (2003)
Major agricultural C3 plants 1000 1500 Larcher (1995)
Radish e 1366 Oh et al. (2015)
Strawberry e 800 da Costa et al. (2014)
82 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

presented in Fig. 6. Characteristically, the shade length during in the interspace was found to be greater with full density
morning (9:00e10:00) and afternoon (15:00e16:00) were higher single row PV array design than with fullþ60% PV density in
than at noon (12:00e13:00). The shade length was higher in double row PV array design and with fullþfullþ60% PV density
triple row PV array (block 3) followed by double row PV array in triple row PV array design. The shaded area on winter sol-
(block 2) and single row PV array (block 1). This was because of stice day as presented above is the maximum possible shade
greater height of PV array in triple row design (2.56 m) than coverage during the year in northern hemisphere. Afterwards
double row (1.96 m) and single row (1.20 m) designs. Shade the solstice, the sun moves northward and thus shade
length in y-axis direction, which was the south-north direc- coverage area is gradually decreased until the summer sol-
tion, was 3.64 m, 2.91 m and 3.14 m in single row PV array stice arrives on 21st June.
(block 1) during morning, noon and afternoon, respectively.
The shade lengths during morning, noon and afternoon were 3.3. Temporal variation of PPFD within AVS
6.12 m, 5.23 m and 5.93 m in block 2 and 8.77 m, 7.55 and
8.51 m in block 3, respectively. Ratio of shade length to the Temporal variation of PPFD in the AVS designs during morn-
total extent of interspace area in y-axis direction was highest ing, noon and afternoon of winter solstice are presented in
during morning at 0.71, 0.52 and 0.47 in single row, double Fig. 7. The area below the PV array received the lowest amount
row, and triple row PV designs, respectively. Alternate of PPFD whereas the interspace area away from the PV
occurrence of light and shade was observed in the shaded area structure received the highest amount of PPFD. At noon, the
of double row and triple row PV array based AVS designs since highest PPFD was received in interspace area of all three PV
the top row of these PV array designs consisted of PV modules array designs as compared to morning and afternoon time.
at 60% density. Variation of solar zenith and azimuth angle Few measurements below the PV array showed significantly
affected the shade length, and therefore shade coverage area higher PPFD than the average values and these were observed
on the ground surface was a dynamic phenomenon. There- at either the east or the west end of the PV array during
fore, low density design of PV array made it possible to receive morning or afternoon, when solar elevation angle was low
both light and shade intermittently where it would otherwise enough to receive direct solar irradiation at these locations.
be completely shaded. Wide variations in measured PPFD in the interspace area of
The shaded areas in the interspace of the AVS during double row and triple row PV array designs were observed.
winter solstice with different PV array designs are presented This was due to 60% PV density of the top row of these PV
in Table 3. Percentage shaded varied from 41.95% to 58.21% of array design, which ultimately led to occurrence of light and
the interspace area of the AVS with single row PV array design shade in alternate patches on the ground surface and the
whereas it varied from 18.01% to 32.40% of interspace area in measured PPFD under shade was lower than the measured
double row PV array design and from 20.14% to 28.59% of PPFD under light. Measured PPFD in shaded area of the
interspace area in triple row PV array design. The shaded area interspace was slightly higher than shaded area below the PV

Fig. 6 e Variation of shade created by ground mounted PV structure in the agrivoltaic system (AVS) at Jodhpur, India during
winter solstice period of 2018 (gray-filled shape represents shade of PV structure).
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 83

Table 3 e Shaded area in the agrivoltaic system (AVS) at Jodhpur, India at different times in a day during winter solstice.
Agrivoltaic system Time Total Shaded Shaded area Shaded area Shaded area Percentage
model area (m2) below PV module in interspace outside block shaded area
(m2) (m2) (m2) in interspace (%)
Single row PV array (Block 1) 9:00e10:00 75.4 19.5 55.9 0.0 58.2
(Total area ¼ 144 m2; 12:00e13:00 66.2 26.0 40.3 e 42.0
Interspace area ¼ 96 m2) 15:00e16:00 73.4 20.1 52.5 0.1 54.7
Double row PV array (Block 9:00e10:00 117.9 62.2 53.8 2.0 32.4
2) 12:00e13:00 103.3 68.7 34.6 e 18.0
(Total area ¼ 288 m2; 15:00e16:00 114.8 63.4 48.7 2.8 25.3
Interspace area ¼ 192 m2)
Triple row PV array (Block 3) 9:00e10:00 193.50 104.5 82.3 6.7 28.6
(Total area ¼ 416 m2; 12:00e13:00 169.47 111.5 58.0 e 20.1
Interspace area ¼ 288 m2) 15:00e16:00 188.41 103.7 77.3 7.4 26.8

array. Likewise, measured PPFD in lit area of intermittent The semivariogram plots show an increasing trend of g with
light/shade zone of the AVS was slightly lower than the PPFD increase in h up to lag distance equals to ‘a’ beyond when it
in lit areas of open space in the AVS. This was because of becomes flat. The average value of ‘a’ across all measurement
scattering of light by the PV array structure which led to the periods was 6.39 m in single row PV array (block 1), 8.87 m in
presence of diffuse radiation in the shaded areas with alter- double row PV array (block 2) and 13.86 m in triple row PV
nate light and shade. This clearly showed that keeping space array (block 3). These range parameters were almost equal to
between two PV modules in the PV array design, by lowering the extent of the AVS designs in y-axis direction, which were
the PV density, provided slightly higher PPFD than the full 4.5 m, 9 m and 12 m in single row, double row and triple row
density PV array design. Therefore, the PV array design with PV array designs, respectively. Thus, the range parameter ‘a’
low density of PV module in AVS is important for successful showed that spatial correlation of PPFD existed within the
cultivation of crops in the interspace area of AVS for which spatial domain of AVS designs. Therefore, PPFD at any un-
PPFD is an essential requirement. measured locations within the extent of AVS designs can be
Descriptive statistics of PPFD in the AVS are presented in estimated using three semivariogram parameters; nugget (C0),
Table 4. Measured PPFD values below PV array area of full partial sill (C) and range (a).
density single row PV array design (block 1) were slightly lower Variation of shade was a dominant factor governing the
as compared to those of double row PV array (block 2) and availability of PPFD. Apart from it, quality of shade also affected
triple row PV array (block 3) designs. This was due to low the availability of PPFD. Therefore, shade covered area alone
height of PV mounting structure in single row PV array design could not fully describe the spatial pattern of PPFD and quality
which led to low chance of getting scattered light below the PV of shade also needs to be considered for mapping spatial
module. Availability of PPFD in partial shaded areas in the pattern of PPFD in AVS designs. For example, measured PPFD in
interspace of double row and triple PV array designs varied the shaded area below the PV array was slightly lower than
widely from as low as 90 mmol m2 d1 to as high as PPFD in shaded area in the interspace of the AVS although both
1280 mmol m2 d1 and the maximum value of this variation areas were observed as shaded areas during the measurement
was almost the same as the available PPFD in open space areas period (Fig. 9). This was due to occurrence of slightly deeper
in the interspace (refers to no shade area in the Table 4). The shade below the PV array than in the interspace area. Lightness
lowest availability of PPFD in the interspace area of the AVS or deepness of shade depended on available scattered light or
was observed during morning and afternoon and these values diffuse short wave radiation in the shaded area. In the inter-
were 194, 100 and 90 mmol m2 d1 in single row, double row, space areas away from the PV array structure, availability of
and triple row PV designs, respectively. Highest availability of scattered light was high and hence the shade was light, which
PPFD in interspace of the AVS was observed during noon and ultimately led to higher availability of PPFD in the shaded area
these were 1209, 1295 and 1287 mmol m2 d1 in single row, of the interspace of the AVS. The availability of PPFD was
double row, and triple row PV designs, respectively. further increased with increase in height of the ground-
mounted PV structure of the AVS. Similar the availability of
3.4. Spatial maps of PPFD in AVS PPFD in the open space (no shade) area is greater than in the lit
area of the interspace of the AVS. Availability of PPFD in the
Semivariogram parameters [nugget (C0), partial sill (C) and open space area (no shade) was greater than the availability of
range (a)] describing spatial variation of PPFD in the AVS are PPFD in the interspace areas that were not covered by shade.
presented in Table 5. C0 and C together described the amount This indicates an increasing gradient of PPFD away from PV
of spatial variation of PPFD. C0 was zero in most cases which mounting structure. These observations confirm that spatial
indicated very low variation at micro scale as well as the low variation of shade was governed by intermittence of shade and
measurement error. It indirectly shows that the variation of light as well as by the height of PV mounting structure.
PPFD in the AVS designs followed a spatial trend, which is also Decreasing the PV density in the AVS designs led to intermit-
evident from the well-defined semivariogram plots (Fig. 8). tence of light and shade and thus increased the availability of
84 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

Fig. 7 e Spatial and temporal variation of PPFD during 20e24 December, 2018 (winter solstice) in the agrivoltaic systems
(AVS) at Jodhpur, India (for a particular distance in the South-North direction, there are 11 data locations across West-East
direction; black dashed line represents the boundary line between below PV array and interspace area and black dotted line
represents imaginary boundary line between partial shade and no shade area in interspace).

PPFD even in shaded areas. Increase in height of PV mounting maps show a range of available PPFD from 0 to
structures also led to increased availability of PPFD. Therefore, 800 mmol m2 s1 during morning and afternoon and from 0 to
the PV density and height of PV mounting structure may be 1400 mmol m2 s1 at around noon. The interspace areas of the
considered as critical factors in the design of AVS, which again PV array designs show a wide variation in PPFD. The PPFD
depends on the site characteristics, e.g. wind regimes, dust values were slightly higher under shade of a single PV module
deposition potential etc. than in the shade of four PV modules in a group (middle
The kriged maps of PPFD, prepared using the spatial vari- portion of block 2 and 3) or two PV modules in a group (two
ation parameters (C, C0 and a), are presented in Fig. 10. These flags ends of block 2 and 3). These observations suggest
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 85

Table 4 e Measured PPFD below PV array and in interspace area within the agrivoltaic system (AVS) at Jodhpur, India
during winter solstice.
Agrivoltaic Time PPFD below PV PPFD in interspace (mmol m2 s1) PPFD in total
system array (mmol m2 s1) system (mmol m2 s1)
Partial shade No shade
Single row PV 9:00e10:00 (Morning) 75 (29e433) 226 (194e444) 426 (394e466) 217 (29e466)
array (Block 1) 12:00e13:00 (Noon) 135 (42e940) 1112 (1050e1150) 1164 (1030e1209) 641 (42e1209)
15:00e16:00 (Afternoon) 79 (33e309) 263 (204e451) 413 (258e467) 221 (33e467)
Double row PV 9:00e10:00 (Morning) 108 (37e547) 264 (100e572) 483 (170e612) 300 (37e612)
array (Block 2) 12:00e13:00 (Noon) 108 (53e235) 644 (134e1190) 1199 (1060e1295) 693 (53e1295)
15:00e16:00 (Afternoon) 130 (41e551) 268 (103e575) 472 (180e630) 304 (41e630)
Triple row PV 9:00e10:00 (Morning) 96 (24e569) 270 (93e608) 521 (195e620) 304 (24e620)
array (Block 3) 12:00e13:00 (Noon) 106 (39e284) 698 (135e1280) 1215 (1150e1287) 706 (39e1287)
15:00e16:00 (Afternoon) 92 (27e541) 241 (90e547) 440 (174e538) 265 (27e538)

Table 5 e Semivariogram parameters of the spatial variation of photosynthetic photon flux density (PPFD) in the agrivoltaic
system (AVS) at Jodhpur, India.
Agrivoltaic system Time Model Nugget Partial Sill Range (m)
Single row PV array (Block 1) 9:00e10:00 Spherical 0 36,385 6.39
12:00e13:00 Spherical 0 347,127 6.63
15:00e16:00 Spherical 0 31,048 6.15
Double row PV array (Block 2) 9:00e10:00 Spherical 1724 46,534 9.09
12:00e13:00 Spherical 0 303,306 8.74
15:00e16:00 Spherical 1622 39,622 8.78
Triple row PV array (Block 3) 9:00e10:00 Spherical 0 55,068 13.66
12:00e13:00 Spherical 0 322,286 12.75
15:00e16:00 Spherical 0 42,328 15.19

categorisation of partial shade area in interspace since the Pearson correlation coefficient (r) between observed and pre-
quality of shade in terms of availability of PPFD in partial dicted PPFD was found to be close to 1 (0.89e0.97) across AVS
shade area differed from the shade below PV array. Moreover, designs, which indicates a strong match between observed and
shade at a particular location of the partial shaded area of predicted values. Prediction accuracy was slightly better in
interspace did not occur throughout the day but was for a few triple row PV array (block 3) than double row PV array (block 2)
hours in a day because the shaded area changed with solar and single row PV array (block 1). A greater number of mea-
zenith angle. surement locations in the triple row PV array might be a reason
The spatial maps of PPFD in the AVS designs may help to for higher prediction accuracy. Similarly, the LCCC statistics
identify the specific crops that could be grown in the inter- were close to 1 in most cases, which indicated strong positive
space area, because the availability of PPFD decides the net agreement between observed and predicted values. The RMSE
rate of photosynthesis in plants. Higher availability of PPFD statistics showed the average deviation of prediction from its
will lead to a greater net rate of photosynthesis, and this is observed values and it was slightly greater during noon than
limited by two critical PPFD levels, the lower is the light during morning or afternoon. This was due to wider variation
compensation point, and the higher is the light saturation in measured values of PPFD around noon. Predicted PPFD in the
point. The PPFD maps in Fig. 10 shows that the available PPFD single row PV array design were slightly negatively biased
during morning to afternoon varied from 0 to 200 mmol m2 s1 whereas in double row and triple row PV array designs, these
below the PV array, from 200 to 600 mmol m2 s1 in the partial were slightly positively biased.
shade area of interspace and 400e1400 mmol m2 s1 in the Observed and predicted PPFDs for different AVS designs
unshaded area of the interspace. Therefore, three cropping and at different times in a day are presented in Fig. 11, which
zones may be delineated in the AVS based on the availability shows points scattered close to along 1:1 line except for those
of PPFD; (i) low light requiring plants for below PV array areas points lying in the middle portion of each plot. The PPFD
(PPFD ¼ 0e200 mmol m2 s1), (ii) moderate light requiring corresponding to these scatter points represented those lo-
plants for interspace areas receiving partial shade (PPFD ¼ cations in AVS where alternate light and shade was observed
200e600 mmol m2 s1) and (iii) high light requiring plants for (partial shaded area) and thus good spatial trend could not be
interspace areas with no shade (PPFD ¼ 400e1400 obtained at these points through semivariogram analysis.
mmol m2 s1). Inclusion of shade coverage map as covariate in kriging pro-
Accuracy of the kriged maps of PPFD was evaluated through cess might have further improved the prediction accuracy of
cross-validation statistics, which are presented in Table 6. PPFD at partial shade area of AVS.
86 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

Fig. 8 e Semivariograms of photosynthetic photon flux density (PPFD) in PV array designs of the agrivoltaic system (AVS) at
Jodhpur, India.

3.5. DLI map in AVS mentioned are greenhouse crops. Future efforts should be
carried out to determine the DLI requirement of field crops
Spatial maps of DLI in the AVS designs are presented in Fig. 12. and accordingly suitable crops may be identified for AVS.
DLI indicates the cumulated daily availability of PAR whereas The DLI map of AVS showed in Fig. 12 corresponds to the
PPFD indicates instantaneous availability of PAR. Since PPFD winter solstice period in northern hemisphere. Therefore,
widely varies in day from morning to afternoon, therefore DLI these DLI maps can only be used to select winter crops (rabi).
may be a good criterion to select crops on the basis of avail- For selection of crops for year-round cultivation in AVS, DLI
ability of light since it quantifies the total amount of light maps for three other critical days, e.g. vernal equinox (23rd
available in a day. Fig 12 shows three distinctive zones of DLI March), summer solstice (21st June) and autumnal equinox
in the AVS designs and these are (i) 0e5 mol m2 d1, (ii) (23rd September), also need to be quantified.
5e20 mol m2 d1 and (iii) 20e30 mol m2 d1. These zones The DLI maps shown in Fig. 12 correspond to AVS with
corresponded to below PV array, interspace with partial shade typical ground mounted PV structure close to ground (~0.5 m
and interspace with no shade, respectively. above ground surface). However, other approaches to design
Based on the availability of DLI in the AVS, suitable crops AVS have also been evolved recently, and these may facilitate a
can be selected. However, information on DLI requirement of wide range of machine deployment within the system. For
crops is limited in the literature (Table 7). Most of these crops example, Trommsdorff et al. (2021) has reported a design of
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 87

Fig. 9 e Availability of PPFD as affected by shade quality (a) Variation of PPFD in shaded areas with distance from PV
structure (b) average PPFD values under different types of shades. (For interpretation of the references to color in this figure
legend, the reader is referred to the Web version of this article.)

Fig. 10 e Spatial maps on availability of PPFD within the agrivoltaic system (AVS) at Jodhpur, India at different times in a day
during winter solstice (red dotted line indicates the boundary of the shade covered area in the AVS and white coloured plus
symbol indicates the location of measurements of PPFD). (For interpretation of the references to color in this figure legend,
the reader is referred to the Web version of this article.)

Table 6 e Cross validation statistics of kriging prediction of PPFD in the agrivoltaic system (AVS) at Jodhpur, India.
Agrivoltaic system Time r LCCC RMSE Bias
Single row PV array (Block 1) 9:00e10:00 0.96 0.95 46.74 0.99
12:00e13:00 0.91 0.89 217.92 15.51
15:00e16:00 0.89 0.87 73.47 10.98
Double row PV array (Block 2) 9:00e10:00 0.93 0.92 76.81 0.82
12:00e13:00 0.95 0.94 153.57 2.81
15:00e16:00 0.92 0.91 75.43 0.64
Triple row PV array (Block 3) 9:00e10:00 0.95 0.95 62.21 4.45  105
12:00e13:00 0.95 0.95 153.30 3.76
15:00e16:00 0.97 0.97 44.01 0.41
88 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

Fig. 11 e Observed and predicted PPFD (mmol m¡2 s¡1) in the agrivoltaic system (AVS) at Jodhpur, India.

AVS in which height of the PV structure above ground surface the standard ground mounted AVS. Reduction of DLI as influ-
was kept 5 m. This not only allowed easy operations of farm enced by ground-mounted PV array although reduces the
machinery within the AVS, but also provided better light dis- availability of photosynthetically active radiation but positively
tribution within the system. The only difficulty with such a influence the field water and energy balance. Limited exposure
high ground-mounted PV structure is the difficulty in manual of ground surface to open sun radiation in the AVS not only
cleaning of deposited dust from the top of PV module. Another reduces the evapotranspiration losses but also reduces the
innovative design of AVS has been reported by Younas et al. ground heat flux. Thus, shade by the PV structure in AVS
(2019) in which vertical bifacial PV modules with east/west suitably modifies the soil hydrothermal regime and improves
facing were used. Bifacial PV modules aligned vertically on the the efficiency of applied irrigation water. For example, Marou
ground surface could provide a much better distribution of et al. (2013b) reported a reduction of actual evapotranspira-
solar irradiation within the AVS as compared to tilted PV tion by 10e30% in AVS where light availability was 50e70% of
module in conventional AVS. It is to be noted here that vertical the full sun irradiation. However, for further increase in water
AVS with bifacial PV module produced almost similar energy use efficiency of crops in AVS system, there is a need to select
output as obtained from tilted AVS with mono facial PV module suitable crops for which light requirement matches with the
(Riaz et al., 2021). However, the PV array density in bifacial PV available DLI in the system and it should be crops with a low
module based AVS was half or even lower relative to that for water requirement with surface cover characteristics.
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 89

Fig. 12 e Daily light integral (DLI) within the agrivoltaic system (AVS) at Jodhpur, India during winter solstice.

Table 7 e DLI requirement of crops.


Crop DLI (mol m2 d1) Reference
Tomato 4e30 Dorais (2003)
Tomato (Green house) 30e35 Spaargaren (2001)
Lettuce (hydroponic) 14e16 Morgan (2013)
Bermuda grass 14.5e44 Glenn et al. (2014)
Lettuce 8e14 Baumbauer et al. (2019)
Cucumber 20e25 Moe et al., 2006, Grimstad (2000).
Strawberry 13.57-45.05 Pc et al. (2019)
Brassica spp 13.9e18.6 Tarakanov and Wang (2009)

average was 5.2 kWh m2 d1 with a variation from


4. Conclusion 3.7 kWh m2 d1 in December to 7.3 kWh m2 d1 in
May). Thus the availability of estimated amount of
AVS of 105 kWp capacity with three PV array designs was PPFD was also quite high throughout the year. Peak
developed at Jodhpur, India. Since, AVS provides the option value of PPFD was generally observed between 12:00
to integrate agriculture with PV generation, availability of and 13:00 and ranged from 1206 mmol m2 s1 in
sufficient amount of PPFD in the interspace as well as below December to 1970 mmol m2 s1 in May. Thus,
the PV array of AVS needs to be investigated. In this study, availability of PPFD was not a constraint in Jodhpur
spatial and temporal variation of shade created by PV ar- for meeting the requirement of photosynthesis for
rays in the AVS and its effect on availability of PPFD and DLI optimum plant growth and yield.
was quantified. Following conclusions are noted from the (ii) However, the ground-mounted PV array with a
study: ground clearance of 0.5 m at the bottom end of the
array created shades not only the areas below the PV
(i) The experimental site of the AVS located at Jodhpur array but also the interspace areas of the AVS. Thus,
in western Rajasthan, India received abundant the shade was expected to decrease the amount of
amount of solar energy throughout the year (annual PPFD in the AVS, which ultimately would affect plant
90 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

growth and yield. Areas below the PV array remained kind help in preparing the engineering drawings of the
under shade during most part in a day throughout manuscript using AutoCAD.
the year. Even, significant amount of shade was
observed in the interspace area during winter sol-
stice. The shaded area in the interspace during the
references
winter solstice varied from 18 to 58% of the total
interspace area. The proportion of shaded area was
Adeh, E. H., Good, S. P., Calaf, M., & Higgins, C. W. (2019). Solar PV
greater in the single row PV array design than in the
power potential is greatest over croplands. Scientific Reports,
double row PV array and triple row PV array designs
9(1), 1e6. https://doi.org/10.1038/s41598-019-47803-3
of the AVS. Adeh, E. H., Selker, J. S., & Higgins, C. W. (2018). Remarkable
(iii) Assessment of spatial maps of PPFD and DLI corre- agrivoltaic influence on soil moisture, micrometeorology and
sponding to winter solstice in Jodhpur, India water-use efficiency. PloS One, 13(11). https://doi.org/10.1371/
revealed three different zones in the AVS for opti- journal.pone.0203256
mising agricultural production. These are below PV Al-Saidi, M., & Lahham, N. (2019). Solar energy farming as a
development innovation for vulnerable water basins.
array, interspace with partial shade and interspace
Development in Practice, 29(5), 619e634. https://doi.org/10.1080/
with no shade. Available amount of PPFD in these 09614524.2019.1600659
three zones were 0e200, 200e600 and 400e1400 mmol Amaducci, S., Yin, X., & Colauzzi, M. (2018). Agrivoltaic systems to
m2 s1, respectively. The DLI in these three zones optimise land use for electric energy production. Applied
were 0e5, 5e20 and 20e30 mol m2 d1, respectively. Energy, 220, 545e561. https://doi.org/10.1016/
These values of available PPFD and DLI are the j.apenergy.2018.03.081
deciding factors for best performance of crops in the Baligar, V. C., Bunce, J. A., Bailey, B. A., Machado, R. C., &
Pomella, A. W. V. (2005). Carbon dioxide and photosynthetic
AVS. Thus, future efforts may be focused to identify
photon flux density effects on growth and mineral uptake of
optimum requirement of PPFD and DLI for different cacao. Journal of Food Agriculture and Environment, 3, 142e147.
field crops as well as high value horticultural crops www.world-food.net.
commonly grown in hot arid region of India. More- Barron-Gafford, G. A., Pavao-Zuckerman, M. A., Minor, R. L.,
over, such DLI maps for other time periods in a year Sutter, L. F., Barnett-Moreno, I., Blackett, D. T., Thompson, M.,
may be calculated e.g. summer solstice, autumnal Dimond, K., Gerlak, A. K., Nabhan, G. P., & Macknick, J. E.
(2019). Agrivoltaics provide mutual benefits across the
equinox and vernal equinox. These DLI maps corre-
foodeenergyewater nexus in drylands. Nature Sustainability,
sponding to four critical days in a year may help to
2(9), 848e855. https://doi.org/10.1038/s41893-019-0364-5
calculate the area and the duration having DLI Baumbauer, D. A., Schmidt, C. B., & Burgess, M. H. (2019). Leaf
exceeding a minimum value of DLI required to grow lettuce yield is more sensitive to low daily light integral than
a particular crop. This will not only help to select kale and spinach. Hort Science, 54(12), 2159e2162. https://
suitable crops for AVS but also to achieve maximum doi.org/10.21273/hortsci14288-19
yield from AVS. Bellow, J. G., & Nair, P. K. R. (2003). Comparing common methods
for assessing understory light availability in shaded-perennial
agroforestry systems. Agricultural and Forest Meteorology,
114(3e4), 197e211.
Castellano, S., Santamaria, P., & Serio, F. (2016). Photosynthetic
Declaration of competing interest photon flux density distribution inside photovoltaic
greenhouses, numerical simulation, and experimental results.
The authors declare that they have no known competing Applied Engineering in Agriculture, 32(6), 861e869. https://doi:10.
financial interests or personal relationships that could have 13031/aea.32.11544@2016.
Chauhan, S. K., Dhillon, W. S., Singh, N., & Sharma, R. (2013).
appeared to influence the work reported in this paper.
Physiological behaviour and yield evaluation of agronomic
crops under agri-horti-silviculture system. International Journal
of Plant Research, 3(1), 1e8. http://doi:10.5923/j.plant.20130301.
01.
Acknowledgments
Chen, N., Gan, Y., & Wang, G. (2003). Photosynthetic responses of
muskmelon (Cucumis melo L.) to photon flux density, leaf
Authors express their sincere thanks to DDG (NRM), Indian temperature and CO2 concentration. Canadian Journal of Plant
Council of Agricultural Research (ICAR) for providing neces- Science, 83(2), 393e399. https://doi.org/10.4141/P02-124
sary financial support to carry out the present study under the Cossu, M., Ledda, L., Urracci, G., Sirigu, A., Cossu, A., Murgia, L.,
ICAR (NRM Division) extramural project titled “Solar farming: Pazzona, A., & Yano, A. (2017). An algorithm for the
Crop production and electricity generation from a single land calculation of the light distribution in photovoltaic
greenhouses. Solar Energy, 141, 38e48. https://doi.org/10.1016/
use system (F. No. NRM 11(16)/2015-AFC (25) dated 10/03/2016
j.solener.2016.11.024
and CAZRI/T-08/EF/114)”. We also express our sincere thanks
da Costa, R. C., Calvete, E. O., Mendonça, H. F. C., & Cecatto, A. P.
to Dr. J.S. Samra and Dr. A.K. Sikka for providing necessary (2014). Phenology, phyllochron, and gas exchanges in frigo and
guidance and encouragements to carry out the research work fresh strawberry ('Fragaria x ananassa'Duch.) plants of cv.
in agrivoltaic system. Credits are also due to all team members Albion. Australian Journal of Crop Science, 8(6), 901.
and staffs, who helped to establish the agrivoltaic system at Dayananda, S. R. J. S. B. (2018). Techno economics feasibility study on
ICAR-Central Arid Zone Research Institute, Jodhpur. Authors agrivoltaic electricity generation in Sri Lanka (Doctoral dissertation).
express their sincere thanks to Mr Chittabrata Santra for his
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 91

Dinesh, H., & Pearce, J. M. (2016). The potential of agrivoltaic Kim, B., Kim, C., Han, S., Bae, J., & Jung, J. (2020). Is it a good time
systems. Renewable and Sustainable Energy Reviews, 54, 299e308. to develop commercial photovoltaic systems on farmland? An
http://doi:10.1016/j.rser.2015.10.024. American-style option with crop price risk. Renewable and
Dorais, M. (2003). The use of supplemental lighting for vegetable Sustainable Energy Reviews, 125, 109827. https://doi.org/10.1016/
crop production: Light intensity, crop response, nutrition, crop j.rser.2020.109827
management, cultural practices. In Canadian greenhouse Kla€ ring, H. P., & Krumbein, A. (2013). The effect of constraining
conference, Toronto, Ontario (Vol. 9). the intensity of solar radiation on the photosynthesis, growth,
Dos Santos, C. N. L. (2020). Agrivoltaic system: A possible synergy yield and product quality of tomato. Journal of Agronomy and
between agriculture and solar energy. Master thesis report. Crop Science, 199(5), 351e359. http://doi:10.1111/jac.12018.
Sweden: KTH Industrial Engineering and Management. Larcher, W. (1995). Physiological plant ecology (3rd ed.). Berlin:
Duffie, J. A., & Beckman, W. A. (2013). Solar engineering of thermal Springer.
processes (4th ed.). Wiley Publication. https://doi.org/10.1002/ Leon, A., & Ishihara, K. N. (2018). Influence of allocation methods
9781118671603 on the LC-CO2 emission of an agrivoltaic system. Resources,
Dupraz, C., Marrou, H., Talbot, G., Dufour, L., Nogier, A., & Conservation and Recycling, 138, 110e117. https://doi.org/
Ferard, Y. (2011). Combining solar photovoltaic panels and 10.1016/j.resconrec.2018.06.017
food crops for optimising land use: Towards new agrivoltaic Lin, L. I. (1989). A concordance correlation coefficient to evaluate
schemes. Renewable Energy, 36(10), 2725e2732. http://doi:10. reproducibility. Biometrics, 45, 255e268.
1016/j.renene.2011.03.005. Li, C., Wang, H., Miao, H., & Ye, B. (2017). The economic and social
Elamri, Y., Cheviron, B., Lopez, J. M., Dejean, C., & Belaud, G. performance of integrated photovoltaic and agricultural
(2018). Water budget and crop modelling for agrivoltaic greenhouses systems: Case study in China. Applied Energy, 190,
systems: Application to irrigated lettuces. Agricultural Water 204e212. https://doi.org/10.1016/j.apenergy.2016.12.121
Management, 208, 440e453. https://doi.org/10.1016/ Loots, N. (2018). Technologic, biological and economic analysis of a
j.agwat.2018.07.001 dynamic agrivoltaic system in the Dutch agriculture sector. Master's
Faust, J. E., Holcombe, V., Rajapakse, N. C., & Layne, D. R. (2005). thesis https://dspace.library.uu.nl/handle/1874/371631.
The effect of daily light integral on bedding plant growth and Majumdar, D., & Pasqualetti, M. J. (2018). Dual use of
flowering. Hort Science, 40(3), 645e649. https://doi.org/10.21273/ agricultural land: Introducing ‘agrivoltaics’ in phoenix
HORTSCI.40.3.645 metropolitan statistical area, USA. Landscape and Urban
Faust, J. E., & Logan, J. (2018). Daily light integral: A research Planning, 170, 150e168. https://doi.org/10.1016/
review and high-resolution maps of the United States. Hort j.landurbplan.2017.10.011
Science, 53(9), 1250e1257. https://doi.org/10.21273/ Malu, P. R., Sharma, U. S., & Pearce, J. M. (2017). Agrivoltaic
HORTSCI13144-18 potential on grape farms in India. Sustainable Energy
Gao, Y., Duan, A., Qiu, X., Sun, J., Zhang, J., Liu, H., & Wang, H. Technologies and Assessments, 23, 104e110. https://doi.org/
(2010). Distribution and use efficiency of photosynthetically 10.1016/j.seta.2017.08.004
active radiation in strip intercropping of maize and soybean. Marrou, H. (2019). Co-locating food and energy. Nature
Agronomy Journal, 102(4), 1149e1157. http://doi:10.2134/ Sustainability, 2(9), 793e794. https://doi.org/10.1038/s41893-
agronj2009.0409. 019-0377-0
Glenn, B., Kruse, J., & Unruh, J. B. (2014). Effect of mowing height Marrou, H., Guilioni, L., Dufour, L., Dupraz, C., & We ry, J. (2013a).
on DLI requirement for warm-season turfgrass. In Microclimate under agrivoltaic systems: Is crop growth rate
C. S. S. A. ASA (Ed.), SSSA international annual meeting. Long affected in the partial shade of solar panels? Agricultural and
Beach, CA. 2e5 Nov. 2014, Paper No 287-1. Forest Meteorology, 177, 117e132. https://doi.org/10.1016/
Goetzberger, A., & Zastrow, A. (1982). On the coexistence of solar- j.agrformet.2013.04.012
energy conversion and plant cultivation. International Journal of Marrou, H., Dufour, L., & Wery, J. (2013b). How does a shelter of
Solar Energy, 1(1), 55e69. solar panels influence water flows in a soilecrop system?
Grimstad, S. (2000). Year round production of cucumber- Better European Journal of Agronomy, 50, 38e51. https://doi.org/
utilisation of production factors. NJF-seminar nr (p. 319). 10.1016/j.eja.2013.05.004
Guertal, E. A., & Elkins, C. B. (1996). Spatial variability of Marrou, H., We ry, J., Dufour, L., & Dupraz, C. (2013c). Productivity
photosynthetically active radiation in a greenhouse. Journal of and radiation use efficiency of lettuces grown in the partial
the American Society for Horticultural Science, 121(2), 321e325. shade of photovoltaic panels. European Journal of Agronomy, 44,
http://doi:10.21273/JASHS.121.2.321. 54e66. https://doi.org/10.1016/j.eja.2012.08.003
Hidaka, K., Dan, K., Imamura, H., Miyoshi, Y., Takayama, T., Marucci, A., & Cappuccini, A. (2016). Dynamic photovoltaic
Sameshima, K., Kitano, M., & Okimura, M. (2013). Effect of greenhouse: Energy efficiency in clear sky conditions. Applied
supplemental lighting from different light sources on growth Energy, 170, 362e376. https://doi.org/10.1016/
and yield of strawberry. Environmental Control in Biology, 51(1), j.apenergy.2016.02.138
41e47. https://doi.org/10.2525/ecb.51.41 Marucci, A., Zambon, I., Colantoni, A., & Monarca, D. (2018). A
Hodges, B. P., Baldwin, C. M., Stewart, B., Tomaso-Peterson, M., combination of agricultural and energy purposes: Evaluation
McCurdy, J. D., Blythe, E. K., & Philley, H. W. (2016). of a prototype of photovoltaic greenhouse tunnel. Renewable
Quantifying a daily light integral for establishment of warm- and Sustainable Energy Reviews, 82, 1178e1186. https://doi.org/
season cultivars on putting greens. Crop Science, 56(5), 10.1016/j.rser.2017.09.029
2818e2826. http://doi:10.2135/cropsci2015.11.0682. Meloni, S., & Sinoquet, H. (1997). Assessment of the spatial
Irie, N., Kawahara, N., & Esteves, A. M. (2019). Sector-wide social distribution of light transmitted below young trees in an
impact scoping of agrivoltaic systems: A case study in Japan. agroforestry system. Annales des sciences forestieres, INRA/EDP
Renewable Energy, 139, 1463e1476. https://doi.org/10.1016/ Sciences, 54(4), 313e333.
j.renene.2019.02.048 Moe, R., Grimstad, S. O., & Gislerod, H. R. (2006). The use of
Kadowaki, M., Yano, A., Ishizu, F., Tanaka, T., & Noda, S. (2012). artificial light in year round production of greenhouse crops in
Effects of greenhouse photovoltaic array shading on Welsh Norway. In V international symposium on artificial lighting in
onion growth. Biosystems Engineering, 111(3), 290e297. http:// horticulture 711 (pp. 35e42).
doi:10.1016/j.biosystemseng.2011.12.006.
92 b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3

Moretti, S., & Marucci, A. (2019). A photovoltaic greenhouse with production and photovoltaic-based electricity generation from
passive variation in shading by fixed horizontal PV Panels. a single land unit. Indian Farming, 68(1), 20e23.
Energies, 12(17), 3269. https://doi.org/10.3390/en12173269 Schindele, S., Trommsdorff, M., Schlaak, A., Obergfell, T.,
Morgan, L. (2013). Daily Light Integral (DLI) and greenhouse tomato Bopp, G., Reise, C., Braun, C., Weselek, A., Bauerle, A., Ho € gy, P.,
production (pp. 10e15). The Tomato Magazine. https://www. & Goetzberger, A. (2020). Implementation of
specmeters.com/assets/1/7/2013_-_DLI_Greenhouse_ agrophotovoltaics: Techno-economic analysis of the price-
Tomato1.pdf. performance ratio and its policy implications. Applied Energy,
Ohashi-Kaneko, K., Takase, M., Kon, N., Fujiwara, K., & 265, 114737. https://doi.org/10.1016/j.apenergy.2020.114737
Kurata, K. (2007). Effect of light quality on growth and Sekiyama, T. (2019). Performance of agrivoltaic systems for shade-
vegetable quality in leaf lettuce, spinach and komatsuna. intolerant crops: Land for both food and clean energy production
Environmental Control in Biology, 45(3), 189e198. https:// (doctoral dissertation). http://nrs.harvard.edu/urn-3:HUL.
doi.org/10.2525/ecb.45.189 InstRepos:42004145.
Oh, S., Moon, K. H., Song, E. Y., Son, I. C., & Koh, S. C. (2015). Sekiyama, T., & Nagashima, A. (2019). Solar sharing for both food
Photosynthesis of Chinese cabbage and radish in response to and clean energy production: Performance of agrivoltaic
rising leaf temperature during spring. Horticulture, systems for corn, a typical shade-intolerant crop.
Environment, and Biotechnology, 56(2), 159e166. https://doi.org/ Environments, 6(6), 65. http://doi:10.3390/
10.1007/s13580-015-0122-1 environments6060065.
Oleskewicz, K. (2020). The Effect of gap spacing between solar panel Sivakumar, M. V. K., & Virmani, S. M. (1984). Crop productivity in
clusters on crop biomass yields, nutrients, and the microenvironment relation to interception of photosynthetically active radiation.
in a dual-use agrivoltaic system. Masters Theses (p. 885). https:// Agricultural and Forest Meteorology, 31(2), 131e141.
scholarworks.umass.edu/masters_theses_2/885. Spaargaren, J. J. (2001). Supplemental lighting for greenhouse crop.
Othman, N. F., Su, A. M., & Ya’acob, M. E. (2018). May. Promising Hortilux schreder B.V. And P.L. Light system, inc., Netherlands.
potentials of agrivoltaic systems for the development of Sun, J., Lu, N., Xu, H., Maruo, T., & Guo, S. (2016). Root zone cooling
Malaysia green economy. IOP Conference Series: Earth and and exogenous spermidine root-pretreatment promoting
Environmental Science IOP Publishing (Vol. 146)(1), Article 012002. Lactuca sativa L. growth and photosynthesis in the high-
http://doi:10.1088/1755-1315/146/1/012002. temperature season. Frontiers in Plant Science, 7, 368. https://
Patel, U. R., & Chauhan, P. M. (2018). Studies of climatic doi.org/10.3389/fpls.2016.00368
parameters under agrivoltaic structure. Renewable and Suthaparan, A., Solhaug, K. A., Stensvand, A., & Gislerød, H. R.
Sustainable Energy: An International Journal (RSEJ), 1(1). (2017). Daily light integral and day light quality: Potentials and
Pc, C. B., Vargas-Cruz, J., Quintero-Arias, G., & Jf, C. (2019). Effect pitfalls of nighttime UV treatments on cucumber powdery
of protected environments on the postharvest quality of mildew. Journal of Photochemistry and Photobiology B: Biology,
strawberry (Fragaria X Ananassa) produced in the tropical 175, 141e148. https://doi.org/10.1016/j.jphotobiol.2017.08.041
mountain areas. International Journal of Biological Tarakanov, I. G., & Wang, J. (2009). Light trophic and signal roles
Instrumentation, 2(1). https://doi.org/10.35840/2631-5025/5104 in the control of morphogenesis of the Brassica plants
Pebesma, Z. E. (2003). Gstat user's manual. P.O. Box 80.115, 3508 TC, developing storage roots. Russian Journal of Plant Physiology,
Utrecht, The Netherlands: Dept. of Physical Geography, 56(2), 232e241. https://doi.org/10.1134/S1021443709020125
Utrecht University. http://www.gstat.org/gstat.pdf. Trommsdorff, M., Kang, J., Reise, C., Schindele, S., Bopp, G.,
Poorter, H., Niinemets, Ü., Ntagkas, N., Siebenka € s, A., Ehmann, A., Weselek, A., Ho € gy, P., & Obergfell, T. (2021).
Ma € enpa
€a€ , M., Matsubara, S., & Pons, T. (2019). A meta-analysis Combining food and energy production: Design of an
of plant responses to light intensity for 70 traits ranging from agrivoltaic system applied in arable and vegetable farming in
molecules to whole plant performance. New Phytologist, 223(3), Germany. Renewable and Sustainable Energy Reviews, 140,
1073e1105. http://doi:10.1111/nph.15754. 110694. https://doi.org/10.1016/j.rser.2020.110694
R Core Team. (2019). R: A language and environment for statistical Valle, B., Simonneau, T., Sourd, F., Pechier, P., Hamard, P.,
computing. Vienna, Austria: R Foundation for Statistical Frisson, T., Ryckewaert, M., & Christophe, A. (2017). Increasing
Computing. https://www.R-project.org/. the total productivity of a land by combining mobile
Ravi, S., Macknick, J., Lobell, D., Field, C., Ganesan, K., Jain, R., photovoltaic panels and food crops. Applied energy, 206,
Elchinger, M., & Stoltenberg, B. (2016). Colocation 1495e1507. https://doi.org/10.1016/j.apenergy.2017.09.113
opportunities for large solar infrastructures and agriculture in Wang, D., & Sun, Y. (2018). Optimizing light environment of the
drylands. Applied Energy, 165, 383e392. https://doi.org/10.1016/ oblique single-axis tracking agrivoltaic system. IOP Conference
j.apenergy.2015.12.078 Series: Earth and Environmental Science. IOP Publishing (Vol. 170)(4),
Riaz, M. H., Imran, H., Younas, R., Alam, M. A., & Butt, N. Z. (2021). Article 042069. http://doi:10.1088/1755-1315/170/4/042069.
Module technology for agrivoltaics: Vertical bifacial versus Wang, D., Sun, Y., Lin, Y., & Gao, Y. (2017a). June. Analysis of light
tilted monofacial farms. IEEE Journal of Photovoltaics, 11(2), environment under solar panels and crop layout. In 2017 IEEE
469e477. https://doi.org/10.1109/JPHOTOV.2020.3048225 44th photovoltaic specialist conference (PVSC), IEEE (pp.
Santra, P., Pande, P. C., Kumar, S., Mishra, D., & Singh, R. K. (2017). 2048e2053).
Agri-voltaics or solar farming: The concept of integrating solar Wang, T., Wu, G., Chen, J., Cui, P., Chen, Z., Yan, Y., Zhang, Y.,
PV based electricity generation and crop production in a single Li, M., Niu, D., Li, B., & Chen, H. (2017b). Integration of solar
land use system. International Journal of Renewable Energy technology to modern greenhouse in China: Current status,
Research (IJRER), 7(2), 694e699. challenges and prospect. Renewable and Sustainable Energy
Santra, P., Singh, R. K., Jain, D., & Yadav, O. P. (2018b). Agri-voltaic Reviews, 70, 1178e1188. https://doi.org/10.1016/
system to increase land productivity and income. Indian j.rser.2016.12.020
Farming, 68(9), 108e111. Wang, D., Zhang, Y., & Sun, Y. (2018). A criterion of crop selection
Santra, P., Singh, R. K., Meena, H. M., Kumawat, R. N., Mishra, D., based on the novel concept of an agrivoltaic unit and m-
Jain, D., & Yadav, O. P. (2018a). Agri-voltaic system: Crop matrix for agrivoltaic systems. In 2018 IEEE 7th world conference
on photovoltaic energy conversion (WCPEC), A joint conference of
b i o s y s t e m s e n g i n e e r i n g 2 0 9 ( 2 0 2 1 ) 7 4 e9 3 93

45th IEEE PVSC, 28th PVSEC & 34th EU PVSEC, IEEE (pp. Yano, A., Kadowaki, M., Furue, A., Tamaki, N., Tanaka, T.,
1491e1496). Hiraki, E., Kato, Y., Ishizu, F., & Noda, S. (2010). Shading and
Webster, R., & Oliver, M. A. (2007). Geostatistics for environmental electrical features of a photovoltaic array mounted inside the
scientists (2nd ed.). Chichester: Wiley. roof of an eastewest oriented greenhouse. Biosystems
Weselek, A., Ehmann, A., Zikeli, S., Lewandowski, I., Schindele, S., Engineering, 106(4), 367e377. http://doi:10.1016/j.
& Ho € gy, P. (2019). Agrophotovoltaic systems: Applications, biosystemseng.2010.04.007.
challenges, and opportunities. A review. Agronomy for Younas, R., Imran, H., Riaz, M. H., & Butt, N. Z. (2019). Agrivoltaic
Sustainable Development, 39(4), 35. https://doi.org/10.1007/ farm design: Vertical bifacial vs. tilted monofacial photovoltaic
s13593-019-0581-3 panels. arXiv preprint arXiv:1910.01076.
Yano, A., & Cossu, M. (2019). Energy sustainable greenhouse crop Zhu, X. G., Long, S. P., & Ort, D. R. (2008). What is the maximum
cultivation using photovoltaic technologies. Renewable and efficiency with which photosynthesis can convert solar energy
Sustainable Energy Reviews, 109, 116e137. https://doi.org/ into biomass? Current Opinion in Biotechnology, 19(2), 153e159.
10.1016/j.rser.2019.04.026 http://doi:10.1016/j.copbio.2008.02.004.

You might also like