Jump to content

This week's Long Read: PV or Trees (and lots of numbers)


Recommended Posts

Available here as the formulas don't come out well. 

https://academic.oup.com/pnasnexus/article/2/11/pgad352/7429362?login=false

 

Photovoltaic fields largely outperform afforestation efficiency in global climate change mitigation strategies 

Rafael Stern, Jonathan D Muller, Eyal Rotenberg, Madi Amer, Lior Segev, Dan Yakir 
PNAS Nexus, Volume 2, Issue 11, November 2023, pgad352, https://doi.org/10.1093/pnasnexus/pgad352
Published:
 
21 November 2023
 Article history
 

Abstract

Suppression of carbon emissions through photovoltaic (PV) energy and carbon sequestration through afforestation provides complementary climate change mitigation (CCM) strategies. However, a quantification of the “break-even time” (BET) required to offset the warming impacts of the reduced surface reflectivity of incoming solar radiation (albedo effect) is needed, though seldom accounted for in CCM strategies. Here, we quantify the CCM potential of PV fields and afforestation, considering atmospheric carbon reductions, solar panel life cycle analysis (LCA), surface energy balance, and land area required across different climatic zones, with a focus on drylands, which offer the main remaining land area reserves for forestation aiming climate change mitigation (Rohatyn S, Yakir D, Rotenberg E, Carmel Y. Limited climate change mitigation potential through forestation of the vast dryland regions. 2022. Science 377:1436–1439). Results indicate a BET of PV fields of ∼2.5 years but >50× longer for dryland afforestation, even though the latter is more efficient at surface heat dissipation and local surface cooling. Furthermore, PV is ∼100× more efficient in atmospheric carbon mitigation. While the relative efficiency of afforestation compared with PV fields significantly increases in more mesic climates, PV field BET is still ∼20× faster than in afforestation, and land area required greatly exceeds availability for tree planting in a sufficient scale. Although this analysis focusing purely on the climatic radiative forcing perspective quantified an unambiguous advantage for the PV strategy over afforestation, both approaches must be combined and complementary, depending on climate zone, since forests provide crucial ecosystem, climate regulation, and even social services.

Significance Statement

Forests and photovoltaic (PV) fields both have significant potential for mitigating climate change, either through carbon uptake by photosynthesis or replacing emissions in energy production. However, both increase global heat load by darkening the surface, and a comparison of the time required to offset this warming impact is critical. Here, we quantify the climate change mitigation potential of both approaches, considering reduced atmospheric carbon, surface energy balance, and land area required to lower global carbon emissions across different climatic zones. Results show that in drylands, PV fields are >50× more efficient than afforestation. In more humid climates, afforestation advantage increases but demands more land than available. Nevertheless, forests provide many other ecological benefits compared with this pure climatic perspective.

Introduction

Maintaining the average global temperature increase below the 1.5°C target which will maintain the Earth system boundaries at a stable state which will remain supporting human civilization (1, 2) demands a substantial and immediate reduction of carbon emissions in the energy sector (3, 4). Low-carbon photovoltaic (PV) electricity generation can significantly contribute to decrease anthropogenic carbon emissions (5, 6), but it requires large-scale infrastructures and is ultimately limited in capacity. Forest restoration (through afforestation of currently unforested land and reforestation of depleted forests) can play an important complementary role in climate change mitigation (CCM) efforts by providing an immediate mean for actively removing atmospheric carbon, which is considered an effective form of long-term carbon storage (3, 7–9). However, both large-scale PV electricity generation and afforestation projects have a high land area demand, which may lead to competition for land in areas that can support forests.

Drylands represent a large proportion of the Earth’s land surface, of which nearly half is semiarid (about 18% of the global land surface area, or about 30 Mkm2) and has attracted attention for potential afforestation projects, due to the availability of large nonforested areas (7, 10–13). However, they also have the potential to host a significant proportion of utility-scale PV fields, taking advantage of the low cloud cover, high insolation at low latitudes, low annual variation in day length, and much lower land-use competition with agriculture, cities, and industries (12–15). The alternative of local PV generation to supply each country's energy demand (rather than concentrating it in drylands) would require hundreds of thousands of square kilometers, greater than their entire urban areas and croplands in some cases, which is unfeasible in many countries (15, 16).

Covering typically bright and reflective drylands with dark PV panels (albedo, α: ∼0.05) (17), or dark forests (α: ∼0.10) (18, 19), can significantly increase surface temperature (Ts) and, in turn, convective heating of the atmospheric boundary layer (20–24). This could counteract the benefits of reduced carbon emissions through PV electricity production (25) or carbon absorption through biomass accumulation in forests (9). Such albedo changes dramatically affect the surface energy budget (Eq. 1):

�n=�in(1−�)+�in−�out=�+LE+�+PVe+�,
(1)

where the net radiation at the surface (Rn) represents the balance of the short-wave incoming solar radiation (Sin) and the outgoing reflected part due to surface albedo, α, and the incoming thermal radiation emitted and reflected by the atmosphere to the surface (Lin) and that emitted and reflected by the surface (Lout). In the case of PV fields, some of the Sin is removed through conversion of energy to electricity, requiring an additional term, PVe (equivalent to Sin × PVeff, i.e. the PV field's land efficiency; see Materials and methods). The surface energy load, represented by Rn, is, in turn, dissipated by nonradiative fluxes, including sensible heat, H, latent heat, LE, some energy storage G on short time scales, and a small fraction used for biochemistry, B (Eq. 1).

In the low LE conditions of water-scarce drylands, the large heat load, enhanced by the albedo changes associated with afforestation or PV fields, contributes to atmospheric warming mainly via heat dissipation through H, which depends on the surface-to-air temperature difference (ΔTs-a) and the aerodynamic resistance to heat transfer (rH) (20, 21, 26, 27). Both afforestation and PV field installations can modify the surface rH in different ways that can be estimated based on measurements of H and ΔTs-a according to Eq. 2:

��=��pΔ�s−a/�,
(2)

where ρ is the density (1.06 ± 0.03 kg m−3) and cp is the heat capacity of air (1016 ± 3 J kg−1 K−1).

Notably, albedo-driven positive radiative forcing (RFr) directly impacts Rn and is a one-step change, having a global warming effect. In contrast, the negative RF achieved by a cumulative reduction of the carbon concentration in the atmosphere (RFc) due to carbon sequestration in afforestation is a cumulative term. By suppressing the future emissions of carbon from fossil fuel burning (carbon emission suppression [CES]), PV electricity generation results in a net reduction in atmospheric carbon when compared with the business-as-usual future scenario. The number of years, n, required for RFc to balance RFr can be estimated as n = RFr/RFc and is termed the break-even time (BET; see Materials and methods).

Previous research on the effects of PV on climate using models and measurements (Table S1) inconclusively predicted an air temperature (Ta) increase (23, 28, 29) or decrease (27, 30, 31). An H increase was effectively shown by measurements in Broadbent et al. (29), and surface temperature (Ts) has been shown to decrease following PV deployment in deserts (32, 33).

Here, we aim to provide a comparative assessment of the BET of land-use change from open drylands to PV fields and afforestation, and an analysis of global climate zones, based on direct field measurements and a detailed analysis of the combined geophysical changes (in α, thermal radiation, H, and rH) and biogeochemical (carbon) climatic effects. This analysis also allows us to compare the land-based efficiency of PV and afforestation approaches as well as their effects on Ts and the potential effects on boundary layer dynamics. Combined, these comparisons can provide critical information for decision-making with respect to CCM by two of the major strategies available today.

Results and discussion

Surface energy fluxes

At our dryland site (see Materials and methods), installation of a PV field led to a ∼45% decrease in surface albedo (α: 0.17 ± 0.04 annual value integrated over the entire PV field; see Materials and methods) compared with the desert (α: 0.38 ± 0.02 annual value; see Materials and methods). The change in albedo (Δα) was 0.16 ± 0.04 in the desert-to-forest transition and 0.21 ± 0.04 in the desert-to-PV field transition, in spite of the differences in absolute albedo values at the two sites (0.28 ± 0.03 and 0.38 ± 0.02 in the background annual values, respectively, due to an increased presence of shrubs and other soil properties outside the forest; Figure S1, Table 3).

Surface heating/cooling is revealed directly by Ts, and indirectly by Lout. We observed only a small (3 W m−2) and insignificant (P = 0.46) mean Lout difference between PV field and desert during summer midday hours (Table 1 and Figure 1). Accordingly, Ts was also very similar in the two sites (Table 1 and Figure 1) due to the similar Lout and emissivity (0.85 and 0.87 in the PV field and the desert, respectively; Methods S2). The difference in Lout between the PV field and desert was also not significant in the other seasons measured in this study: 11 W m−2 (P = 0.69) and 12 W m−2 (P = 0.9) in autumn and spring, respectively. In the Yatir afforestation, both Ts and Lout were significantly lower (P < 0.001) than in the surrounding desert by ∼7°C and 45 W m−2, respectively (Table 1). In spring, Lout difference between the afforestation and its desert background was 38 W m−2 (P < 0.01), and Ts difference was ∼5°C, although P = 0.96 (Table S4). No measurements were performed with the mobile station in the afforestation desert background in autumn. For the purpose of the RFr calculation, an annual value of ∼10% of the ΔLout of 18 W m−2 between the afforestation and the desert is considered to escape directly to space and participate in RF (34, 35) (Table 2; see Materials and methods). Compared with the practically unchanged Ts in the desert-to-PV field transition, this highlights the forest's efficiency in heat dissipation and its capability to decrease Ts. In other words, the forest has a local surface cooling effect, while the PV field installation has no effect on Ts during midday hours.

Fig. 1.

Land-use change effects. Desert-to-afforestation (A, B) and desert-to-PV (C, D) transitions in drylands, during summer midday. The size of the arrows represents the magnitude, with straight and round arrows for radiative and nonradiative fluxes, respectively. The green arrow represents carbon sequestration (photosynthesis) or suppression of emissions (represented by the thermoelectric power plant) in the afforestation and PV field, respectively. The thermometer represents the local cooling effect on surface temperature that the forest has.

 

     
         
         
         
         
         
         
         
         
         

 

 
Table 1.

Geophysical effects of the deployment of PV field or forest in open drylands.

  PV Afforestation
Land use PV field Background Forest Background
Rn (W m−2) 529 (47) 410 (35) 650 (89) 452 (71)
H (W m−2) 358 (76) 199 (50) 522 (97) 225 (51)
LE (W m−2) 35 (28) 29 (25) 43 (48) 33 (42)
Lout (W m−2) 583 (16) 586 (26) 505 (15) 550 (17)
Ts (°C) 49.5 (2.2) 49.6 (3.7) 37.1 (2.5) 43.9 (2.5)
Ta (°C) 35.7 (2.6) 37.2 (3.0) 30.9 (2.3) 29.3 (1.9)
ΔTs-a (°C) 13.8 (1.1) 12.4 (1.2) 6.3 (1.0) 14.5 (2.1)
rH (s m−1) 44.0 (12) 70 (15) 13 (3) 74 (25)

Values represent summer midday (10 AM–3 PM) means, and SD is in parentheses. Air (Ta, °C) and mean surface temperatures (Ts, °C), surface–air temperature difference (ΔT = Ts  Ta, °C), sensible heat flux (H, W m−2), and aerodynamic resistance to heat transfer (rH, s m−1) are shown for each land-use type.

 

           
           
           
     
           
           
           
     
     
           

 

 
Table 2.

Geophysical and geochemical RF effects and the BET for three fossil fuel scenarios and the world mixture of fossil fuels.

  Coal Oil Gas World mixture Afforestation
CO2F (kgCO2 kWh−1) 0.87 0.67 0.40 0.54 N/A
CF (kgC kWh−1) 0.24 0.18 0.11 0.15 N/A
LCA (kgC kWh−1) 0.026 N/A
CF-LCA (kgC kWh−1) 0.21 0.15 0.08 0.12 N/A
CES or sink (kgC m−2 year−1) 26.0 19.3 10.2 14.9 0.15–0.28
RFc (W m−2 × 10−14) −7.15 −5.31 −2.81 −4.11 −4.12 × 10−2
RFr local (W m−2) 65.66 39.38
RFr global (W m−2 × 10−13) 1.01 0.76
Break-even time (years) 1.4 1.9 3.6 2.5 94–185

CO2F is the CO2 footprint (kg kWh−1), and CF is the carbon footprint (kg kWh−1), obtained from IRENA for fossil fuel scenarios for the year 2005 (the last year before renewables became larger than 1% of the global mixture) (6), and the conversion factor of C(kg) = CO2(kg) × 0.27. LCA of the PV panels is from da Silva Lima et al. (36) (see Materials and methods). CF-LCA is the carbon footprint of the PV panels after the carbon emitted during the panel's life cycle (mining, manufacturing, transport, installation, operation, and disposal or recycling and electricity storage in lithium-ion batteries) is deduced from its mitigation potential. CES is based on this PV field EPa (EPa = 123 kWh m−2 year−1). RFc (W m−2) is the radiation forcing of the carbon component, while RFr (W m−2) is the radiation forcing of the radiation components (albedo and longwave thermal radiation). CO2 and C footprints of fossil fuels and LCA of PV panels are not applicable to afforestation (N/A).

Neither Ta (5 m above ground level) nor ΔTs-a above our PV field was significantly different from the adjacent desert (Table 1; Tables S3 and S4 and Discussion S1), in contrast to previous studies that showed a heat island effect (23, 26, 28, 29, 33) (Table S1). While this could be explained by some intrinsic differences between the sites (e.g. background vegetation removal reducing evaporative cooling and less dense PV installations (28, 33)), a possible underestimation of heat dissipating H fluxes in models is likely an important factor. Indeed, considering large H fluxes increased model precision considerably for the prediction of Ta and Ts related to PV systems in a previous study (37).

An H increase of 25–50 W m−2 was recently reported in a PV field in a desert in Arizona, United States (29). Our results show a more pronounced effect with an H increase of ∼160 W m−2 during summer midday hours in the desert-to-PV transition (Table 1). One of the main differences is that our study focused in the summer season, while Broadbent et al. (29) focused in the spring and under different insolation patterns. In afforestation, the mechanism behind the H increase of 297 W m−2 in the summer (Table 1 and Figure 1), despite the cooler surface, is associated with increased roughness and turbulence that decreases aerodynamic resistance to heat transfer, rH (20, 21, 26, 29), and has previously been termed a “convector effect” (22, 38). Note that the apparent contradiction—of the afforestation presenting half the ΔTs-a of the PV field but ∼1.45× the PV H flux—is explained by an almost 3.4× lower rH (Table 1, Discussion S2).

Different effects of land-use change on Ts, H, and on boundary layer depth (20, 21) can have important implications when considered at a large scale due to their potential impact on local and regional climate, wind convergence, and convection, with potential effects on cloudiness, rain, and temperature (11, 20, 23, 39). The Ts reduction achieved in the afforestation (Table 1 and Figure 1; Tables S3 and S4) should be considered critically, as the heat dissipation to the planetary boundary layer would still yield increased global RF, unless a short BET is achieved.

BET

Based on the calculated CES and the PV life cycle analysis (LCA), we estimated the RFc and, in turn, the BET for the PV installation in drylands, resulting in a range of 1.4–3.6 years for the three different fossil fuel scenarios (Table 2). In comparison, many hydropower reservoirs have a BET of 4–80 years while having an albedo lower than that of PV fields (0.08) (40). In contrast, the annual carbon assimilation rate of dryland forests—∼0.15 kgC m−2 year−1 for Yatir (41) or 0.28 kgC m−2 year−1 for global dryland forest generalization (7)—results in a BET which is up to 2× longer than the expected main forest C accumulation period (e.g. 80 years in Braakhekke et al. (42), Table 2).

A further comparison between PV fields and afforestation reveals stark differences in the land area required to offset carbon emissions from electricity production. The global electricity demand estimated for 2050 (6) is 55,000 TWh year−1, which would result in an emission of ∼6.6 GtC year−1, considering the global fossil fuel mix (Table 2; see Materials and methods). Sequestering this emission by dryland afforestation would require ∼26–44 Mkm2 (based on Qubaja et al. (41), or the estimated mean for global drylands (7), respectively), significantly more than the estimated 4 Mkm2 available for dryland afforestation (43). Emission suppression using PV would require a mere 0.44 Mkm2 in the same climate zone (0.75% of global drylands (12), Table 3), without considering the extra space needed to deal with production intermittency (15). This is mainly due to a 100× factor of the CES by PV energy versus annual photosynthesis of the nearby afforestation considered in this study (14.9 versus 0.15 kgC m−2 year−1, Tables 2 and 3 and Figure 1), with little difference in the albedo change. While any carbon uptake is considered important, we estimated here a long BET that could limit the land-based efficiency of afforestation as a useful CCM approach compared with PV fields in drylands. Therefore, we emphasize the complementary combination of reducing emissions through PV usage in drylands and further removal of atmospheric C into storage in biomass in the temperate and tropical zones could be a useful combination of CCM strategies. However, this assessment, based on the present study, could be sensitive to variations in climatic conditions (present and future ones), including teleconnection effects and ecohydrological feedbacks on both atmosphere and soil conditions.

 

       
 
       
       
       
       
       
       
       
 
       
       
       
       
       
       
       
       
       
       
       
       
       

 

 
Table 3.

Global perspective of biogeochemical and biogeophysical effects of afforestation and PV field installation.

  Temperate Tropical Drylands
Afforestation
αforest 0.14 0.13 0.12
αbackground 0.20 0.18 0.28
 Δαbackground-forest 0.06 0.05 0.16
 C sequestration (kgC m−2 year−1) 0.31 0.41 0.15–0.28
 Break-even time (years) 18.5 15.4 94–175
 Available land area (Mkm2) 2.1 4.7 4
 Required land area (Mkm2) 21.3 16.1 26–44
PV
αPV field 0.13 0.10 0.17
αbackground 0.20 0.18 0.38
 Δαbackground-PV 0.07 0.08 0.21
 CES (kgC m−2 year−1) 12.1 15.8 14.9
 Break-even time (years) <1 <1 1.4–3.6
 Required area (Mkm2) 0.55 0.42 0.44
 Area (m2) 110,000 130,000 75,600
Eg (kWh m−2 year−1) 900 1,500 2,122
 EP (GWh year−1) 10 17 9.3
 Land fraction of panels (fpv) 0.45 0.60 0.43
 Panel efficiency (PVeff,panels) 0.15 0.14 0.12
 Land efficiency (PVeff) 0.11 0.09 0.06
 EPa (kWh m−2 year−1) 90 130 123

Sites in the temperate (Germany) and tropical (Panama, Methods S3) zones were selected that had both forest and grassland ecosystems. Carbon sequestration by afforestation of temperate and tropical forests was obtained from Luyssaert et al. (44) and for dryland forests from Qubaja et al. (41) and Veldman et al. (9). Albedo (α) was obtained from the data available in Fluxnet and Euroflux networks and from the literature in the case of the tropical α for the forest and grassland (45). The reported values are medians, thus reducing the impact of extreme years. Percentages of PV panels and total areas were estimated from Google Earth (Methods S3). The estimated electricity generation by the PV field (kWh year−1) was obtained from the GPPD v. 1.3 database. EPa (kWh m−2 year−1) was calculated using Eq. M5. Available land for afforestation was obtained from Griscom et al. (8) for temperate and tropical climates and from Rohatyn et al. (43) for drylands. The land areas required to offset 6.6 GtC year−1 are reported for all cases. Land areas necessary for afforestation are sensitive to carbon sequestration, here obtained from Fluxnet sites, but could nearly double using estimates for the entire climate zone by Griscom et al. (8).

Temperate and tropical regions have higher precipitation rates and are therefore more suitable for forest growth than drylands. Afforestation in these zones exhibits a significantly lower BET (15.4–18.5 years, Table 3) than in drylands and, in turn, a smaller land area required to sequester the carbon emissions (16.1–21.3 Mkm2, Table 3). However, this still vastly exceeds the available land area for afforestation in each climate zone, and complete afforestation of the available land in both regions would achieve less than half of the carbon sequestration required (2.6 GtC year−1) (8) to balance anthropogenic fossil fuel emissions. At a global scale, the entire area available for forest restoration would have to be used (18 Mkm2 of forest cover) (7), representing ∼40× more land area than that required by PV in drylands.

PV fields are less sensitive to their location, i.e. neither the BET nor the land area required for CES differs significantly among climate zones (Table 3). The tropical and temperate zones have higher cloud cover and lower solar angles with shorter insolation hours during a significant part of the year. Nevertheless, the resulting reduction in insolation was likely outweighed by the angle and density of PV installation and/or different technologies that increase panel efficiency (Methods S3). Therefore, despite lower annual electricity production (EPa) in the temperate zone, the gross land efficiency for electricity generation of PV fields was highest there (Table 3). Although the BET of PV fields in these regions is very small due to a much lower Δα effect (<1 year, Table 3), installation could lead to further carbon emissions related to impacts on background vegetation (15).

Indeed, PV electricity generation can prevent much higher carbon emissions than the amount that afforestation can sequester. However, the active removal of past emitted carbon and the possible long-term storage as biomass and soil organic matter provides an important advantage for afforestation. Moreover, it is important to consider that our study presents a purely climatic perspective, as forests can provide a variety of additional natural and cultural ecosystem services that are not provided by industrial PV systems (e.g. biodiversity, green space, soil protection, wood production, recreation, tourism, mental and physical health, and other societal needs) (46–48). Hence, these should be considered in policy and management decisions in a broader perspective of CCM plans.

Besides carbon emission, other environmental impacts associated with the installation of PV fields in large ecosystem areas can occur to a different degree, such as soil erosion, dust production, increased risk of fires, water pollution and soil contamination, noise and light pollution, alteration of nutrient dynamics, biodiversity loss, heavy metals released to the environment (48, 49), and even sociological and cultural impacts (46). Nevertheless, since the land required for PV is much smaller than that needed for afforestation in drylands for CCM, the scale of the impacts in these areas would be significantly lower. Note also that some forms of dryland afforestation also have significant biodiversity impacts (50, 51).

In conclusion, we note that the large dryland fraction of Earth's land surface must be considered in developing any large-scale CCM strategy. In this context, drylands offer the potential for both PV and afforestation, and we offer a much-needed comparative, quantitative analysis of both strategies. We show that PV clearly provides greater land-use efficiency in drylands, but some of the advantages of afforestation (e.g. long-term C storage or surface cooling) should not be neglected.

Materials and methods

Measurement location

The research part concerning the PV fields’ impact on drylands was conducted in a PV field in a hyperarid region in the Arava valley in Israel, next to kibbutz Ketura (Figure S1). The mean annual rain is about 28 mm, with an average of 1.3 days of rain events above 5 mm, aridity index < 0.03. The mean annual temperature is 23.9 ± 8°C for the years 2001–2019, with mean daily maximum summer temperatures (July and August) reaching 38.9 ± 1.8°C and a mean daily winter minimum of 9.5 ± 2.8°C (January and February) (Israel Meteorology Service). The annual average of the incoming shortwave radiation is 245 W m−2 (Israel Meteorological Service) or 2,122 kWh m−2 (52).

The PV field chosen has an area of 75,600 m2 and an installed production capacity of 4.9 MW. Rows of monofacial nontracking crystalline silicon PV panels intercalated with soil compose 84% of the field (from which 51% are PV panels and 49% is soil, Figure S3), while 16% of the field area is composed of infrastructure (roads and buildings, measured from satellite images in QGIS software). Considering the whole PV field with the PV panels and soil rows, and the necessary infrastructure, the total coverage of the panels is 43% versus 57% of mainly bare soil. This PV field is about 400 m long and has a maximum width of 250 m. Each panel has an extension of 5 m vertically, 2 m horizontally, and 0.05 m width. All PV panel rows are fixed and facing south. The height between the top of the panel and the ground is 3 m, and the bottom is 0.5 m above ground. The gap between each of the lines of panels is 5 m, and the inclination angle is 30°.

The afforestation data were measured in a fixed eddy covariance (EC) station at the Yatir forest (41) (Figure S1). It is located in a 2,800 ha afforestation of mainly Pinus halepensis trees with a height of ∼10 m and a 50% canopy cover (estimated from Lidar data, June 2019) in the dry southern Mediterranean region, at the northern edge of the Negev desert in Israel (31°20′49″N; 35°03′07″E; altitude 600 to 850 m above sea level; cf. Methods S6) (41, 53). The site contains an EC flux tower operating since 2001, with EC sensors ∼10 m above canopy. Up- and downwelling shortwave (0–4,000 nm, W m−2; CM21, Kipp & Zonen B.V., Delft, The Netherlands) and longwave radiation (4–50 µm, W m−2; Eppley, Newport, RI) were measured at the same tower, ∼5 m above canopy.

Note that the drylands considered in this study (from the semiarid to arid transition zone suitable for forestation or PV and the hyperarid suitable mainly for PV) represent the vast majority of global drylands from the climatological perspective (Figure S1f) and the global range of dryland surface albedo (Methods S1) (54).

Mobile system measurements and data analysis

The mobile system was previously used and described in a number of recent studies (21, 35, 55–61). It consists of a full EC (62) and meteorological station transported by a 12-ton, 4WD truck (Figure S4), which provides air-conditioned instrument facility, internet communication, and a uninterrupted power supply (UPS) system that ensures the energy stability. It was powered by the electricity of a nearby kibbutz (Ketura) during the PV desert background measurements, through an infrastructure building of the Arava Power Company inside the PV field, as well as a generator during measurements in deserts outside the Yatir forest. Sensors include a sonic anemometer (R3-100, Gill Instruments, Lymington, Hampshire, United Kingdom) for 3D wind speed components (horizontal u and v, and the vertical w) and temperature derived from the speed of sound (63), an enclosed-path CO2/H2O Infrared Gas Analyzer (IRGA; LI-7200, LI-COR Lincoln, NE, United States), air temperature, relative humidity (HMP45C probes, Campbell Scientific, Logan, UT, United States), and pressure sensors (Campbell Scientific, Logan, UT, United States). Radiation sensors included the shortwave solar (0.28–2.8 µm, factory calibrated to the full 0 to 4 µm range; CMP21, Kipp&Zonen, Delft, The Netherlands) and longwave or thermal radiation (4.5–42 µm factory calibrated to the full 4–50 µm range; CGR-4, Kipp&Zonen Delft, The Netherlands). All the sensors were placed on top of a pneumatic mast that connects to the truck, which was positioned up to 35 m away from the truck in the center of the field sites with location and height aimed to provide a sufficient “fetch” for the EC system (Figure S5). The mast height was 2.7 m (spring), 4.5 m (autumn), and 5.1 m (summer) in the PV field desert background, 5.2 m (spring), 4.2 m (autumn), and 4.5 m (summer) in the PV field, and 5.0 m (2013) and 4.6 m (2015) at the Yatir desert sites. The measuring height of the sonic anemometer is located 0.63 m above the mast height. Raw EC data were collected using a custom LabView software (developed by Enrico Segre, 2010), and meteorological data were collected by a Campbell Scientific CR3000 logger (Logan, UT, United States). The IRGA was cleaned and calibrated for zero and span values before each season. A metal box deployed between the truck and mast hosts the IRGA's air pump (10 lpm flow) and the data logger for the radiation and the temperature and humidity sensors. More details about the mobile laboratory, with validations and calibrations, can be found in Asaf et al. (55) and Helman et al. (60). Soil heat storage (G) was not measured in this study.

Field measurements were performed in spring (2018 March 18–21 in the desert background and 2018 March 22–27 in the PV field), autumn (2018 October 16–23 in the desert background and 2018 October 24–31 in the PV field), and summer (2019 July 09–15 in the desert background and 2019 July 16–24 in the PV field), therefore covering different meteorological conditions and solar radiation incidence and sun angles. The reported dates correspond to full days with all the half-hour averages included (Table S2). Campaign-based measurements were proven to be a valid methodology in order to provide seasonal comparison between different land uses (21, 35, 55–61). The data for each campaign were continuous, with <3% of half-hourly gaps filled using the Marginal Distribution Sampling tool in Tovi. Comparisons with nearby long-term meteorological stations show that Sin and Ta conditions were well representative of each season (Figure S2).

Sensible (H) and latent heat (LE) flux data collection, processing, and postprocessing were performed according to established methodology (64–66). It included quality control flagging (67), sonic anemometer double rotation tilt corrections (68), and compensation of air density fluctuations due to water vapor variations (69–71), as well as spectral corrections of EC data (72, 73) and footprint and shear velocity (74, 75) estimation (Figure S6). We used the EddyPro 7.0.5 (LICOR Biosciences, Lincoln, Nebraska) and the novel Tovi 2.8.1 software (LICOR Biosciences, Lincoln, Nebraska). No air heat storage correction was applied since it was assumed to be negligible in magnitude. Due to the very open “canopy” characteristic of our measurement sites with the mobile station (PV field and desert backgrounds of the PV field and afforestation), total flux errors derived from storage are expected to be small (76).

EC flux measurements can be susceptible to bias due to instrument location, which we rigorously tested: Firstly, stable structures can result from the module arrangement, which can enhance convection (77). However, an analysis of cospectra provides confidence in our measurements and does not indicate directional effects. Secondly, an estimation of the random error in flux measurements (78) showed a mean error of 10 W m−2 in the desert background and in the PV field for H and 7 W m−2 for LE. Such uncertainties have previously been described for only the highest quality data (67), which adds confidence to our measurements. Such uncertainties can influence the precision of the energy balance estimations but not the conclusions of a clear H enhancement over the PV field. Additional future studies are required to better understand the effect of geometrical arrangements on fluxes.

The effect of shortwave radiation on the longwave radiation sensors was estimated using Eq. M1, obtained by a shading experiment as described in Pérez et al. (79):

�=0.38+0.01���,
( M1)

where Sin is the shortwave radiation measured by the mobile system radiation sensor, and the corrected longwave radiation value is obtained by subtracting ΔL from the raw longwave radiation sensor measurement. This equation is applied both for the incoming (downwelling, Lin) and for the outgoing (upwelling, Lout) radiation components.

Mean seasonal and annual albedo values were calculated as the ratio between incoming (downwelling, Sin) and outgoing (upwelling, Sout) shortwave radiation (�=�out�in)⁠, on a half-hourly basis. PV desert background albedo showed negligible variations throughout the year due to the complete absence of vegetation (spring: 0.40; summer: 0.37; and autumn: 0.38). Annual albedo for the desert background (for RF calculation, see below) was calculated as the mean of all data, containing the combined data sets of all seasons. PV field albedo could not be measured directly due to height restrictions imposed by the field operator that were sufficient for flux data but not for radiative data (Methods S4). For this reason, PV field albedo was obtained through a combination of albedo data of its components and their fraction, as viewed from nadir: We used a set of drone imagery taken over the PV field throughout the day during the summer and autumn campaigns to measure the fraction of PV panels as well as exposed and shaded soil. Then, a cosine curve was fitted to each campaign's shade fraction as a function of time, with an additional set of cosine curves fitted to the yearly course as a function of day of year, in order to obtain the shaded soil fraction for each half-hour of the year. The PV panel fraction was a fixed 51%, and the remaining fraction was exposed soil. A value of 0.05 for PV panel α (17) was combined with the albedo of each time of the day of the desert background α for exposed soil. For shaded soil, the exposed soil α was multiplied by the diffuse radiation fraction (Methods S5) obtained from above-canopy measurements at the nearby Yatir station. Finally, seasonal and annual α were calculated as weighted means by Sin across the seasonal campaigns or the entire year, respectively.

RF and “BET” calculations

The balance between the negative RF due to CES of PV power generation or carbon sequestration by forests (RFc; cumulative over time) and the positive RF due to changes in the surface energy budget of PV fields and forests (RFr; a one-time change) is used to estimate the BET required to balance the two effects (Eq. M2).

�RFc=RFr(Wm−2),
( M2)

where n is the number of years to reach a balance. RFr is defined as follows:

RFr=[Δ��g]+0.1Δ�out�E(Wm−2),
( M3)

where Δ� is the albedo change of the desert-to-PV or desert-to-forest transition. When the local effect is considered, the efficiency of the land area of a PV field in generating electricity (PVeff) additionally needs to be subtracted from Δ� to account for the amount of incoming solar radiation that leaves the field as electricity and therefore reduces the local Rn (net radiation at the surface) (Eq. 1). We note that the land efficiency of a PV field is different from the panel efficiency (Methods S6 and below). Eg is the global radiation of the region (245 W m−2), and ΔLout is the change in upwelling longwave radiation of this land cover change multiplied by the fraction directly emitted to space (∼10% (34)). ΔLout was not significant in the desert-to-PV transition (Results and Discussion S2) and was therefore only relevant in the RFr estimation of afforestation (annual scale: 18 W m−2) (35). AE is the surface area of planet Earth (5.1 × 1014 m2), needed in order to obtain the global effect and to compare with the effect of the RF of CO2 (RFc) (25). The estimation of RFc is based on the update by Etminan et al. (80) of the original equation of Myhre et al. (81) as used in Betts et al. (82):

RFC=[�1(�−�0)2+�1|�−�0|+�1�¯+5.36]ln(��0)(Wm−2)
( M4)

where a1 = −2.4 × 10−7 W m−2 ppm−1, b1 = −7.2 × 10−4 W m−2 ppm−1, c1 = −2.1 × 10−4 W m−2 ppm−1, C0 is the background CO2 concentration of the atmosphere (409.85 ppm on average in 2019), C is the new CO2 mixing ratio after the forcing caused by the land-use change (ΔC + C0), and �¯ is the background N2O mixing ratio of the atmosphere (331 ppb on average 2019). CO2 and N2O average background mixing ratios for the year of 2019 (the year we finished our measurements) were obtained from NOAA/GML (https://gml.noaa.gov/ccgg/trends/). ΔC was obtained from the expression ΔC = E × AF/k, where E is the emission suppression (CES) in the case of PV installations, or carbon sequestration for an afforestation, AF is the carbon airborne fraction (0.46) (3), and k is the conversion from kilograms of carbon (kgC) to parts per million (ppm) (2.16 × 1012) (25). Note that this AF means that with regard to the effects on atmospheric carbon, any addition or removal of C is translated by the AF factor of 0.46 in terms of the atmospheric C pool due to the C partitioning among the atmosphere, land, and oceans. In the case of PV fields, E must be estimated from its EPa reduced by the carbon emitted during a panel's life cycle and energy storage (LCA, see below).

In order to estimate the EPa of a PV field per square meter, we used the following:

EPa=365⋅EPd�pv=365⋅25,50075,600=123kWhm−2year−1,
( M5)

where EPd is the annual mean of the daily electricity production provided by the PV field operator (25,500 kWh day−1) and Apv is the area of the PV field (75,600 m2), multiplied by 365 days per year to obtain an annual estimate. In the case of our PV field, EPa was about 123 kWh m−2 year−1. Since this is the real production that the PV field operator gets as a final product, it already includes losses related to the performance ratio (PR)—which ranges typically between 0.85 and 0.75 (83–86)—and that comprises losses related to the DC–AC current inverter, cables, and transformers (83, 87). Therefore, EPa can be used directly to estimate the efficiency of the PV field. The PR is mainly influenced by temperature (14), and values provided by the operator of our study site fluctuated between >0.85 in the spring and slightly below 0.80 in the summer. Note that the PV panels’ high temperature (almost 60°C during peak, Discussion S1) also affects efficiency, and there is a decrease of ∼0.5% of the efficiency for every degree increase above 25°C (88–90). Nevertheless, PV fields can operate at a much higher temperature than forests, as those are physiologically limited to a maximum of 40–50°C (91) and require water for carbon sequestration and therefore are negatively affected by decreasing precipitation. In contrast, PV fields benefit from increased insolation resulting from lower cloudiness. The annual CES was estimated by multiplying EPa by the annual carbon emission per kilowatt hour produced according to each fossil fuel or mixture of fuels (Table 2).

The CO2 footprint (CO2F, which corresponds to the emission of CO2 for each kilowatt hour produced) for each fuel type was obtained from the IEA (92). The impact of replacing thermoelectric electricity generation with PV generation was analyzed for three different fossil fuel sources—coal, oil, and gas, or a global average mixture of fuels (mainly coal, oil, and natural gas). We used the global average mixture of fuels from the year 2005 because it was just before the beginning of the large-scale deployment of renewable energy sources, weakening the signal of fossil fuel burning replacement. The carbon footprint (CF) was obtained by multiplying the CO2F values by a factor of ∼0.27 (molar mass C/molar mass CO2). The CES of each scenario of fuel replacement by PV electricity production (coal, oil, or gas thermoelectric power plants) is obtained according to Eq. M6:

CES=EPa(CF-LCA),
( M6)

where CES is expressed in units of kgC m−2 year−1, and LCA is the life cycle analysis of PV installations. CES included the following: (i) EPa per surface area of the PV field as a whole, including rows of soil between panels and infrastructure (EPa of ∼123 kWh m−2 year−1 in our study site). This is mostly determined by the efficiency of the entire PV field (PVeff = 5.8%) and includes losses due to the PV PR (i.e. the DC–AC current inverter, losses due to cables and transformers, and temperature-related efficiency decreases (93)) and the field cover parcel of the PV panels (43% in our case). (ii) The CF (i.e. emission of C per kilowatt hour produced) of fossil fuel energy sources that are replaced by PV for three scenarios of electricity production, i.e. either replacing 100% coal or gas production or that of the average global energy mix (92), providing the perspective of replacing mainly fossil fuels before the recent increase in alternative energy use. (iii) The carbon emission of PV electricity generation, based on a LCA that includes emissions from mining, manufacturing, transport, installation, operation, and disposal or recycling of PV panels and storage methods (36) (here considering lithium-ion batteries as the most used form of storage due to its space efficiency and energy density) (36). LCA values for PV electricity are typically in the 0.010–0.050 kgCO2eq kWh−1 range (25, 83, 94–96) before considering storage. The 0.039 kgCO2eq kWh−1 value (94) has a high resemblance to the conditions at our study (PV efficiency = 6% and irradiance 1,935 kWh m−2 year−1). When storage is added, a 0.095 kgCO2 kWh−1 LCA value is obtained (36), corresponding to 0.026 kgC kWh−1. Note that energy storage includes losses of between 5 and 40% in different storage options (36), but due to the high uncertainty and the possibility of demand side management of electricity use, it was assumed to be a minor component and was not included. Therefore, the LCA value of 0.026 kgC kWh−1 was used in this study (36).

Based on the EPa of the PV field of this study (123 kWh m−2 year−1), and on the annual solar irradiance of the region (2,122 kWh m−2), we estimated the mean annual land-use efficiency of this entire PV field (PVeff) at ∼5.8%(=EPa�g=1232,122)⁠. Note that the efficiency of PV panels at this field (PVeff, panel) is comparable to other different locations due to the different panel inclination angles (Methods S3). This value represents the actual ability to convert the incoming radiation to electricity (97) and was the same even when calculated separately for each season. It considers the diverse losses after the electricity generation, including the PR. It is important to note that this efficiency is for the PV field as a landscape unit, with all its different components (PV panels, sun-exposed soil and shaded soil, and infrastructure), because the focus of this study is to compare it with an alternative land use such as afforestation for carbon mitigation purposes. All calculations were done in Python using the numpy (98) and gmpy2 (99) libraries for precision purposes.

Detailed thermal radiation data were obtained through a deep learning segmentation algorithm for the different elements of the PV field (Figure S7 and Methods S4): PV panels, sun-exposed soil, and shade, and combined with ecosystem emissivity (Methods S2). These measurements were compared with data of continuous carbon and energy fluxes at the permanent forest flux measurement site (IL-Yat, http://www.europe-fluxdata.eu/home/sites-list) in the nearby semiarid Yatir pine forest plantation (Figure S1) (41). Desert background data both south (August 2013, March 2014) and north (August 2015, March 2016) of the forest were also obtained using the mobile system. We combined these analyses with two additional climate zones (temperate and tropical) using data obtained from other flux stations, analyses of satellite images, and data gathered from the literature (Methods S3).

Acknowledgments

We thank Amnon Cochavi, Fyodor Tatarinov, Itay Oz, Roni Tal, Yonatan Darom, Guy Sadot, Eitan Ribak, and Revital Weiss for the help in the field campaigns; Efrat Schwartz, Gitanjali Thakur, Irina Vishnevetsky, Boaz Ninyo Setter, Hagay Sagi, Haim Madar, Valentin Zabolotko and their team for the assistance with mechanical and technical tasks; and Matthias Mauder for advice with regard to data analysis. We thank Shani Rohatyn for field measurements in Yatir desert background and essential discussions regarding calculations. We also thank the Arava Institute for Environmental Studies, Kibbutz Ketura, the Arava Power Company, and Tareq Abuhamed for allowing us to perform this study in their photovoltaic field and on their desert land; Christian Bernhofer and J. Thomas Gruenwald for sharing data of the Fluxnet sites DE-Gri and DE-Hzd; and Nina Buchmann, Sebastian Wolf, and Sabina Keller for sharing the data of the Fluxnet sites PA-SPn and PA-SPs. We thank Gerardo Fratini for the assistance with eddy covariance spectral analysis. We thank Ron Milo for his valuable comments on the manuscript and Rafael Fleishman for important clarifications regarding PV panel functioning.

Supplementary Material

Supplementary material is available at PNAS Nexus online.

Funding

This study was supported by Jewish National Fund-KKL (1851), Minerva Foundation (#714147), the Israel Science Foundation (#1976/17), and by a research grant from the Yotam project and the Weizmann Institute Sustainability and Energy Research Initiative, as well as a fellowship from the Society of the Swiss Friends of the Weizmann Institute.

Author Contributions

Conceptualization: R.S., J.D.M., E.R., M.A., and D.Y. Methodology: R.S., J.D.M., E.R., M.A., L.S., and D.Y. Data acquisition: R.S., J.D.M., E.R., M.A., and Visualization: R.S. and J.D.M. Funding acquisition: R.S., J.D.M., and D.Y. Project administration: E.R. and D.Y. Supervision: E.R. and D.Y. Writing—original draft: R.S. and J.D.M. Writing—review and editing: R.S., J.D.M., E.R., M.A. and D.Y.

Data Availability

The data underlying this analysis are freely available from the following sources: flux data for the sites in Israel: https://doi.org/10.34933/403d99aa-17e2-4f67-95a4-5de13681439c; GPPD database (v.1.1.0): http://datasets.wri.org/dataset/globalpowerplantdatabase; Fluxnet (https://fluxnet.org/); and Euroflux (http://www.europe-fluxdata.eu/). The Python scripts used to analyze the data are freely available under https://github.com/rafastern/Dryland_PV_vs_afforestation.

References

 
1

 

Mckay
 
DIA
, et al.  
2022
. 
Exceeding 1.5°C global warming could trigger multiple climate tipping points
. 
Science
 
377
:
eabn7950
.

 

 
 
2

 

Rockström
 
J
, et al.  
2023
. 
Safe and just Earth system boundaries
. 
Nature
 
619
:
102
111
.

 

 
 
3

 

Friedlingstein
 
P
, et al.  
2020
. 
Global carbon budget 2020
. 
Earth Syst Sci Data
. 
12
:
3269
3340
.

 

 
 
4

 

Welsby
 
D
, 
Price
 
J
, 
Pye
 
S
, 
Ekins
 
P
. 
2021
. 
Unextractable fossil fuels in a 1.5°C world
. 
Nature
 
597
:
230
234
.

 

 
 
5

 

Green
 
MA
. 
2019
. 
Photovoltaic technology and visions for the future
. 
Prog Energy
. 
1
:
013001
.

 

 
 
6

 

IRENA
. 
2019
. 
Future of solar photovoltaic: deployment, investment, technology, grid integration and socio-economicaspects (A Global Energy Transformation: paper), International Renewable Energy Agency, Abu Dhabi
.

 

 
7

 

Bastin
 
J-F
, et al.  
2019
. 
The global tree restoration potential
. 
Science
 
365
:
76
79
.

 

 
 
8

 

Griscom
 
BW
, et al.  
2017
. 
Natural climate solutions
. 
Proc Natl Acad Sci U S A
. 
114
:
11645
11650
.

 

 
 
9

 

Veldman
 
JW
, et al.  
2019
. 
Comment on “The global tree restoration potential”
. 
Science
 
366
:
1
5
.

 

 
 
10

 

Goffner
 
D
, 
Sinare
 
H
, 
Gordon
 
LJ
. 
2019
. 
The Great Green Wall for the Sahara and the Sahel Initiative as an opportunity to enhance resilience in Sahelian landscapes and livelihoods
. 
Reg Environ Change
. 
19
:
1417
1428
.

 

 
 
11

 

Yosef
 
G
, et al.  
2018
. 
Large-scale semi-arid afforestation can enhance precipitation and carbon sequestration potential
. 
Sci Rep
. 
8
:
1
10
.

 

 
 
12

 

Bastin
 
J
, et al.  
2017
. 
The extent of forest in dryland biomes
. 
Science
 
356
:
635
638
.

 

 
 
13

 

Rohatyn
 
S
, 
Yakir
 
D
, 
Rotenberg
 
E
, 
Carmel
 
Y
. 
2022
. 
Limited climate change mitigation potential through forestation of the vast dryland regions
. 
Science
 
377
:
1436
1439
.

 

 
 
14

 

Kawajiri
 
K
, 
Oozeki
 
T
, 
Genchi
 
Y
. 
2011
. 
Effect of temperature on PV potential in the world
. 
Environ Sci Technol
. 
45
:
9030
9035
.

 

 
 
15

 

Van De Ven
 
DJ
, et al.  
2021
. 
The potential land requirements and related land use change emissions of solar energy
. 
Sci Rep
. 
11
:
2907
.

 

 
 
16

 

Capellán-Pérez
 
I
, 
de Castro
 
C
, 
Arto
 
I
. 
2017
. 
Assessing vulnerabilities and limits in the transition to renewable energies: land requirements under 100% solar energy scenarios
. 
Renew Sustain Energy Rev
. 
77
:
760
782
.

 

 
 
17

 

Burg
 
BR
, 
Ruch
 
P
, 
Paredes
 
S
, 
Michel
 
B
. 
2015
. 
Placement and efficiency effects on radiative forcing of solar installations
. 
AIP Conf Proc
. 
1679
:
090001-1
090001-9
. https://doi.org/10.1063/1.4931546.

 

 
18

 

Rotenberg
 
E
, 
Yakir
 
D
. 
2011
. 
Distinct patterns of changes in surface energy budget associated with forestation in the semiarid region
. 
Glob Change Biol
. 
17
:
1536
1548
.

 

 
 
19

 

Arora
 
VK
, 
Montenegro
 
A
. 
2011
. 
Small temperature benefits provided by realistic afforestation efforts
. 
Nat Geosci
. 
4
:
514
518
.

 

 
 
20

 

Brugger
 
P
, et al.  
2018
. 
Effect of surface heterogeneity on the boundary-layer height: a case study at a semi-arid forest
. 
Boundary-Layer Meteorol
. 
169
:
233
250
.

 

 
 
21

 

Banerjee
 
T
, et al.  
2018
. 
Turbulent transport of energy across a forest and a semiarid shrubland
. 
Atmos Chem Phys
. 
18
:
10025
10038
.

 

 
 
22

 

Rotenberg
 
E
, 
Yakir
 
D
. 
2010
. 
Contribution of semi-arid forests to the climate system
. 
Science
 
327
:
451
454
.

 

 
 
23

 

Li
 
Y
, et al.  
2018
. 
Climate model shows large-scale wind and solar farms in the Sahara increase rain and vegetation
. 
Science
 
361
:
1019
1022
.

 

 
 
24

 

Lu
 
Z
, et al.  
2020
. 
Impacts of large-scale Sahara solar farms on global climate and vegetation cover
. 
Geophys Res Lett
. 
48
:
1
10
.

 

 
25

 

Nemet
 
GF
. 
2009
. 
Net radiative forcing from widespread deployment of photovoltaics
. 
Environ Sci Technol
. 
43
:
2173
2178
.

 

 
 
26

 

Nguyen
 
KC
, 
Katzfey
 
JJ
, 
Riedl
 
J
, 
Troccoli
 
A
. 
2017
. 
Potential impacts of solar arrays on regional climate and on array efficiency
. 
Int J Climatol
. 
37
:
4053
4064
.

 

 
 
27

 

Hu
 
A
, et al.  
2016
. 
Impact of solar panels on global climate
. 
Nat Clim Change
. 
6
:
290
294
.

 

 
 
28

 

Barron-Gafford
 
GA
, et al.  
2016
. 
The photovoltaic heat island effect: larger solar power plants increase local temperatures
. 
Sci Rep
. 
6
:
1
7
.

 

 
 
29

 

Broadbent
 
AM
, 
Krayenhoff
 
ES
, 
Georgescu
 
M
, 
Sailor
 
DJ
. 
2019
. 
The observed effects of utility-scale photovoltaics on near-surface air temperature and energy balance
. 
J Appl Meteorol Climatol
. 
58
:
989
1006
.

 

 
 
30

 

Salamanca
 
F
, 
Georgescu
 
M
, 
Mahalov
 
A
, 
Moustaoui
 
M
, 
Martilli
 
A
. 
2016
. 
Citywide impacts of cool roof and rooftop solar photovoltaic deployment on near-surface air temperature and cooling energy demand
. 
Boundary-Layer Meteorol
. 
161
:
203
221
.

 

 
 
31

 

Taha
 
H
. 
2013
. 
The potential for air-temperature impact from large-scale deployment of solar photovoltaic arrays in urban areas
. 
Sol Energy
. 
91
:
358
367
.

 

 
 
32

 

Zhang
 
X
, 
Xu
 
M
. 
2020
. 
Assessing the effects of photovoltaic powerplants on surface temperature using remote sensing techniques
. 
Remote Sens
. 
12
:
8
14
.

 

 
33

 

Chang
 
R
, et al.  
2018
. 
Observed surface radiation and temperature impacts from the large-scale deployment of photovoltaics in the barren area of Gonghe, China
. 
Renew Energy
. 
118
:
131
137
.

 

 
 
34

 

Trenberth
 
KE
, 
Fasullo
 
JT
, 
Kiehl
 
J
. 
2009
. 
Earth's global energy budget
. 
Bull Am Meteorol Soc
. 
90
:
311
324
.

 

 
 
35

 

Rohatyn
 
S
, 
Rotenberg
 
E
, 
Tatarinov
 
F
, 
Carmel
 
Y
, 
Yakir
 
D
. 
2023
. 
Large variations in afforestation-related climate cooling and warming effects across short distances
. 
Commun Earth Environ
. 
4
:
18
.

 

 
 
36

 

da Silva Lima
 
L
, et al.  
2021
. 
Life cycle assessment of lithium-ion batteries and vanadium redox flow batteries-based renewable energy storage systems
. 
Sustain Energy Technol Assess
. 
46
:
101286
.

 

 
37

 

Heusinger
 
J
, 
Broadbent
 
AM
, 
Sailor
 
DJ
, 
Georgescu
 
M
. 
2020
. 
Introduction, evaluation and application of an energy balance model for photovoltaic modules
. 
Sol Energy
. 
195
:
382
395
.

 

 
 
38

 

Banerjee
 
T
, 
De Roo
 
F
, 
Mauder
 
M
. 
2017
. 
Explaining the convector effect in canopy turbulence by means of large-eddy simulation
. 
Hydrol Earth Syst Sci
. 
21
:
2987
3000
.

 

 
 
39

 

Branch
 
O
, 
Wulfmeyer
 
V
. 
2019
. 
Deliberate enhancement of rainfall using desert plantations
. 
Proc Natl Acad Sci U S A
. 
116
:
18841
18847
.

 

 
 
40

 

Wohlfahrt
 
G
, 
Tomelleri
 
E
, 
Hammerle
 
A
. 
2021
. 
The albedo–climate penalty of hydropower reservoirs
. 
Nat Energy
. 
6
:
372
377
.

 

 
 
41

 

Qubaja
 
R
, 
Grünzweig
 
JM
, 
Rotenberg
 
E
, 
Yakir
 
D
. 
2020
. 
Evidence for large carbon sink and long residence time in semiarid forests based on 15 year flux and inventory records
. 
Glob Change Biol
. 
26
:
1626
1637
.

 

 
 
42

 

Braakhekke
 
MC
, et al.  
2019
. 
Modeling forest plantations for carbon uptake with the LPJmL dynamic global vegetation model
. 
Earth Syst Dyn
. 
10
:
617
630
.

 

 
 
43

 

Rohatyn
 
S
, 
Rotenberg
 
E
, 
Yakir
 
D
, 
Carmel
 
Y
. 
2021
. 
Assessing climatic benefits from forestation potential in semi-arid lands
. 
Environ Res Lett
. 
16
:
104039
.

 

 
 
44

 

Luyssaert
 
S
, et al.  
2007
. 
CO2 balance of boreal, temperate, and tropical forests derived from a global database
. 
Glob Change Biol
. 
13
:
2509
2537
.

 

 
 
45

 

Science Panel for the Amazon.
 
2021
. Amazon Assessment Report 2021. In: 
Nobre
 
C
, et al. , editors. 
United Nations sustainable development solutions network
. 
New York, USA
.

 

 
 
46

 

Grodsky
 
SM
, 
Hernandez
 
RR
. 
2020
. 
Reduced ecosystem services of desert plants from ground-mounted solar energy development
. 
Nat Sustain
. 
3
:
1036
1043
.

 

 
 
47

 

Orenstein
 
DE
. 
2021
. 
The cultural ecosystem services of Mediterranean pine forests
. 
631
655
.

 

 
48

 

Turney
 
D
, 
Fthenakis
 
V
. 
2011
. 
Environmental impacts from the installation and operation of large-scale solar power plants
. 
Renew Sustain Energy Rev
. 
15
:
3261
3270
.

 

 
 
49

 

Hernandez
 
RR
, et al.  
2014
. 
Environmental impacts of utility-scale solar energy
. 
Renew Sustain Energy Rev
. 
29
:
766
779
.

 

 
 
50

 

Veldman
 
JW
, et al.  
2015
. 
Where tree planting and forest expansion are bad for biodiversity and ecosystem services
. 
Bioscience
 
65
:
1011
1018
.

 

 
 
51

 

Hawlena
 
D
, 
Bouskila
 
A
. 
2006
. 
Land management practices for combating desertification cause species replacement of desert lizards
. 
J Appl Ecol
. 
43
:
701
709
.

 

 
 
52

 

Vardimon
 
R
. 
2011
. 
Assessment of the potential for distributed photovoltaic electricity production in Israel
. 
Renew Energy
. 
36
:
591
594
.

 

 
 
53

 

Muller
 
JD
, et al.  
2021
. 
‘Dual-reference’ method for high-precision infrared measurement of leaf surface temperature under field conditions
. 
New Phytol
. 
232
:
2535
2546
.

 

 
 
54

 

Tsvetsinskaya
 
EA
, 
Schaaf
 
CB
, 
Gao
 
F
, 
Strahler
 
AH
, 
Dickinson
 
RE
. 
2006
. 
Spatial and temporal variability in Moderate Resolution Imaging Spectroradiometer-derived surface albedo over global arid regions
. 
J Geophys Res Atmos
. 
111
:
1
10
.

 

 
 
55

 

Asaf
 
D
, et al.  
2013
. 
Ecosystem photosynthesis inferred from measurements of carbonyl sulphide flux
. 
Nat Geosci
. 
6
:
186
190
.

 

 
 
56

 

Rohatyn
 
S
, et al.  
2018
. 
Differential impacts of land use and precipitation on “ecosystem water yield”
. 
Water Resour Res
. 
54
:
5457
5470
.

 

 
 
57

 

Cochavi
 
A
, et al.  
2021
. 
Differential responses to two heatwave intensities in a Mediterranean citrus orchard are identified by combining measurements of fluorescence and carbonyl sulfide (COS) and CO2 uptake
. 
New Phytol
. 
230
:
1394
.

 

 
 
58

 

Seco
 
R
, et al.  
2017
. 
Springtime ecosystem-scale monoterpene fluxes from Mediterranean pine forests across a precipitation gradient
. 
Agric For Meteorol
. 
237–238
:
150
159
.

 

 
59

 

Yang
 
F
, 
Qubaja
 
R
, 
Tatarinov
 
F
, 
Rotenberg
 
E
, 
Yakir
 
D
. 
2018
. 
Assessing canopy performance using carbonyl sulfide measurements
. 
Glob Change Biol
. 
24
:
3486
3498
.

 

 
 
60

 

Helman
 
D
, et al.  
2017
. 
A biophysical approach using water deficit factor for daily estimations of evapotranspiration and CO2 uptake in Mediterranean environments
. 
Biogeosciences
 
14
:
3909
3926
.

 

 
 
61

 

Eder
 
F
, et al.  
2015
. 
Secondary circulations at a solitary forest surrounded by semi-arid shrubland and their impact on eddy-covariance measurements
. 
Agric For Meteorol
. 
211–212
:
115
127
.

 

 
62

 

Burba
 
G
, 
Chen
 
W
, 
Venables
 
DS
, 
Sigrist
 
MW
, 
Chen
 
W
et al.  
2021
. Atmospheric flux measurements. In: 
Chen
 
W
, 
Venables
 
DS
, 
Sigrist
 
MW
, editors. 
Advances in spectroscopic monitoring of the atmosphere
. 
Amsterdam, Netherlands
: 
Elsevier Science
. p. 
618
.

 

 
63

 

Burba
 
G
, 
Anderson
 
D
. 
2010
. 
A brief practical guide to eddy covariance flux measurements
. https://doi.org/10.1890/06-1336.1.

 

 
64

 

Fratini
 
G
, 
Mauder
 
M
. 
2014
. 
Towards a consistent eddy-covariance processing: an intercomparison of EddyPro and TK3
. 
Atmos Meas Tech
. 
7
:
2273
2281
.

 

 
 
65

 

Foken
 
T
, 
Wichura
 
B
. 
1996
. 
Tools for quality assessment of surface-based flux measurements
. 
Agric For Meteorol
. 
78
:
83
105
.

 

 
 
66

 

Vickers
 
D
, 
Mahrt
 
L
. 
1997
. 
Quality control and flux sampling problems for tower and aircraft data
. 
J Atmos Ocean Technol
. 
14
:
512
526
.

 

 
 
67

 

Mauder
 
M
, 
Foken
 
T
. 
2006
. 
Impact of post-field data processing on eddy covariance flux estimates and energy balance closure
. 
Meteorol Zeitschrift
. 
15
:
597
609
.

 

 
 
68

 

Wilczak
 
JM
, 
Oncley
 
SP
, 
Stage
 
SA
. 
2001
. 
Sonic anemometer tilt correction algorithms
. 
Boundary-Layer Meteorol
. 
99
:
127
150
.

 

 
 
69

 

Webb
 
EK
, 
Pearman
 
GI
, 
Leuning
 
R
. 
1980
. 
Correction of flux measurements for density effects due to heat and water vapour transfer
. 
Q J R Meteorol Soc
. 
106
:
85
100
.

 

 
 
70

 

Burba
 
G
, et al.  
2012
. 
Calculating CO2 and H2O eddy covariance fluxes from an enclosed gas analyzer using an instantaneous mixing ratio
. 
Glob Change Biol
. 
18
:
385
399
.

 

 
 
71

 

Ibrom
 
A
, 
Dellwik
 
E
, 
Larsen
 
SE
, 
Pilegaard
 
K
. 
2007
. 
On the use of the Webb-Pearman-Leuning theory for closed-path eddy correlation measurements
. 
Tellus Ser B Chem Phys Meteorol
. 
59
:
937
946
.

 

 
 
72

 

Moncrieff
 
J
, 
Clement
 
R
, 
Finnigan
 
J
, 
Meyers
 
T
. 
2004
. Averaging, detrending, and filtering of eddy covariance time series. In: 
Lee
 
X
, 
Massman
 
W
, 
Law
 
B
, editors. 
Handbook of micrometeorology
. Atmospheric and oceanographic sciences library, vol 
29
. 
Dordrecht
: 
Springer
. p. 
7
31
. https://doi.org/10.1007/1-4020-2265-4_2.

 

 
 
73

 

Fratini
 
G
, 
Ibrom
 
A
, 
Arriga
 
N
, 
Burba
 
G
, 
Papale
 
D
. 
2012
. 
Relative humidity effects on water vapour fluxes measured with closed-path eddy-covariance systems with short sampling lines
. 
Agric For Meteorol
. 
165
:
53
63
.

 

 
 
74

 

Kljun
 
N
, 
Calanca
 
P
, 
Rotach
 
MW
, 
Schmid
 
HP
. 
2004
. 
A simple parameterisation for flux footprint predictions
. 
Boundary-Layer Meteorol
. 
112
:
503
523
.

 

 
 
75

 

Kljun
 
N
, 
Calanca
 
P
, 
Rotach
 
MW
, 
Schmid
 
HP
. 
2015
. 
A simple two-dimensional parameterisation for Flux Footprint Prediction (FFP)
. 
Geosci Model Dev
. 
8
:
3695
3713
.

 

 
 
76

 

Nicolini
 
G
, et al.  
2018
. 
Impact of CO2 storage flux sampling uncertainty on net ecosystem exchange measured by eddy covariance
. 
Agric For Meteorol
. 
248
:
228
239
.

 

 
 
77

 

Stanislawski
 
B
, 
Margairaz
 
F
, 
Cal
 
RB
, 
Calaf
 
M
. 
2020
. 
Potential of module arrangements to enhance convective cooling in solar photovoltaic arrays
. 
Renew Energy
. 
157
:
851
858
.

 

 
 
78

 

Finkelstein
 
PL
, 
Sims
 
PF
. 
2001
. 
Sampling error in eddy correlation flux measurements
. 
J Geophys Res Atmos
. 
106
:
3503
3509
.

 

 
 
79

 

Pérez
 
M
, 
Alados-Arboledas
 
L
. 
1999
. 
Effects of natural ventilation and solar radiation on the performance of pyrgeometers
. 
J Atmos Ocean Technol
. 
16
:
174
180
.

 

 
 
80

 

Etminan
 
M
, 
Myhre
 
G
, 
Highwood
 
EJ
, 
Shine
 
KP
. 
2016
. 
Radiative forcing of carbon dioxide, methane, and nitrous oxide: a significant revision of the methane radiative forcing
. 
Geophys Res Lett
. 
43
:
12614
12623
.

 

 
 
81

 

Myhre
 
G
, 
Highwood
 
EJ
, 
Shine
 
KP
, 
Stordal
 
F
. 
1998
. 
New estimates of radiative forcing due to well mixed greenhouse gases
. 
Geophys Res Lett
. 
25
:
2715
2718
.

 

 
 
82

 

Betts
 
RA
. 
2000
. 
Offset of the potential carbon sink from boreal forestation by decreases in surface albedo
. 
Nature
 
408
:
187
190
.

 

 
 
83

 

Peng
 
J
, 
Lu
 
L
, 
Yang
 
H
. 
2013
. 
Review on life cycle assessment of energy payback and greenhouse gas emission of solar photovoltaic systems
. 
Renew Sustain Energy Rev
. 
19
:
255
274
.

 

 
 
84

 

Fthenakis
 
VM
, 
Hyung
 
CK
, 
Alsema
 
E
. 
2008
. 
Emissions from photovoltaic life cycles
. 
Environ Sci Technol
. 
42
:
2168
2174
.

 

 
 
85

 

Alsema
 
EA
, 
de Wild-Scholten
 
MJ
, 
Fthenakis
 
VM
. 
2006
. 
Environmental impacts of PV electricity generation-a critical comparison of energy supply options. In 21st European Photovoltaic Solar Energy Conference, EU-PVSEC Dresden. p. 3201–3207
. https://www.ctc-n.org/sites/www.ctc-n.org/files/resources/21-eupvsc-alsema-dewild-fthenakis.pdf.

 

 
86

 

Jahn
 
U
, 
Nasse
 
W
. 
2004
. 
Operational performance of grid-connected PV systems on buildings in Germany
. 
Prog Photovolt Res Appl
. 
12
:
441
448
.

 

 
 
87

 

Fthenakis
 
V
, et al.  
2011
. 
Methodology guidelines on life cycle assessment of photovoltaic electricity
. 3rd ed.

 

 
88

 

Al- Sabounchi
 
AM
. 
1998
. 
Effect of ambient temperature on the demanded energy of solar cells at different inclinations
. 
Renew Energy
. 
14
:
149
155
.

 

 
 
89

 

Kaldellis
 
JK
, 
Kapsali
 
M
, 
Kavadias
 
KA
. 
2014
. 
Temperature and wind speed impact on the efficiency of PV installations. Experience obtained from outdoor measurements in Greece
. 
Renew Energy
. 
66
:
612
624
.

 

 
 
90

 

Skoplaki
 
E
, 
Palyvos
 
JA
. 
2009
. 
On the temperature dependence of photovoltaic module electrical performance: a review of efficiency/power correlations
. 
Sol Energy
. 
83
:
614
624
.

 

 
 
91

 

Muller
 
JD
, 
Rotenberg
 
E
, 
Tatarinov
 
F
, 
Oz
 
I
, 
Yakir
 
D
. 
2021
. 
Evidence for efficient nonevaporative leaf-to-air heat dissipation in a pine forest under drought conditions
. 
New Phytol
. 
232
:
2254
2266
.

 

 
 
92

 

IEA
. 
2019
. 
CO2 emissions from fuel combustion 2019
. 
Paris, France
: 
OECD
.

 

 
93

 

Skoplaki
 
E
, 
Palyvos
 
JA
. 
2009
. 
Operating temperature of photovoltaic modules: a survey of pertinent correlations
. 
Renew Energy
. 
34
:
23
29
.

 

 
 
94

 

Meier
 
PJ
, 
Kulcinski
 
GL
. 
2002
. 
Life-cycle energy requirements and greenhouse gas emissions for building-integrated photovoltaics
. Fusion Technology Institute, University of Wisconsin, Madison, USA.

 

 
95

 

Ito
 
M
, 
Kato
 
K
, 
Komoto
 
K
, 
Kichimi
 
T
, 
Kurokawa
 
K
. 
2008
. 
A comparative study on cost and life-cycle analysis for 100 MW very large-scale PV (VLS-PV) systems in deserts using m-Si, a-Si, CdTe, and CIS modules
. 
Prog Photovolt Res Appl
. 
16
:
17
30
.

 

 
 
96

 

Fthenakis
 
VM
, 
Kim
 
HC
. 
2011
. 
Photovoltaics: life-cycle analyses
. 
Sol Energy
. 
85
:
1609
1628
.

 

 
 
97

 

Scherba
 
A
, 
Sailor
 
DJ
, 
Rosenstiel
 
TN
, 
Wamser
 
CC
. 
2011
. 
Modeling impacts of roof reflectivity, integrated photovoltaic panels and green roof systems on sensible heat flux into the urban environment
. 
Build Environ
. 
46
:
2542
2551
.

 

 
 
98

 

Harris
 
CR
, et al.  
2020
. 
Array programming with NumPy
. 
Nature
 
585
:
357
362
.

 

 
 
99

 

Van Horsen
. 
2023
. 
gmpy2: gmpy2 interface to {GMP}/{MPIR}, {MPFR}, and {MPC} for Python 2.7 and 3.5+ (August 7, 2023)
. https://pypi.org/project/gmpy2/.

 

 

Author notes

1

R.S. and J.D.M. contributed equally to this work.

Competing Interest: The authors declare no competing interest.

© The Author(s) 2023. Published by Oxford University Press on behalf of National Academy of Sciences.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
 
 
 

Supplementary data

Link to comment
Share on other sites

Sorry couldn't be bothered reading it, but why either or, can't you have both. We are on a hill site, with lots of mature trees, we have a roof array with optimisers, but take a hit due to shade in the morning. The other array ground mounted to miss most shading.

Link to comment
Share on other sites

Just now, JohnMo said:

Sorry couldn't be bothered reading it, but why either or, can't you have both. We are on a hill site, with lots of mature trees, we have a roof array with optimisers, but take a hit due to shade in the morning. The other array ground mounted to miss most shading.

The main reason is that PV can reduce CO2 emissions (compared to natural gas) in 2 to 5 years, trees take 50 years.

Rewilding desert regions can be a bit tricky as water supplies need to be diverted, and that can take away from food production.

 

But of course we should be doing both, but not in a binary way i.e. if we can't have trees or PV, then we will do nothing.

 

The fastest way to reduce deforestation is manage forestation and agriculture better, but that will take a lot of education.

  • Like 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...