Minimizing environmental impacts of solar farms: a review of current science on landscape hydrology and guidance on stormwater management

Given that solar farms involve major construction activities during the site development process, there is potential for them to have impacts on soil properties and vegetation. Soil physical and chemical properties directly impact hydrologic processes, and thus we begin our review with understanding these impacts.

Some physical, chemical, and biological soil indicators can be lower on solar farms compared to semi-natural land cover types, depending on solar farm management strategies. Research on a solar farm in southern France with silty clay soils did not find differences in soil bulk density on solar farms as compared to semi-natural pinelands and shrublands, but did find a reduced water holding capacity (Lambert et al 2021). Research on a solar farm in Colorado, USA observed a greater coarse particle fraction on the solar farm as compared to an adjacent native grassland reference. The reason for the difference in particle size is likely the soil disturbance and vegetation removal during the construction phase of the solar farm, which causes erosion of the fine particles (Choi et al 2020a).

Some studies have found the carbon and nitrogen content to be lower in the soil on solar farms than in reference soils (Choi et al 2020a, Lambert et al 2021). Basal respiration and microbial biomass have also been measured at lower quantities on solar farms compared to reference land covers (Lambert et al 2021). However, at a solar farm on reclaimed cropland with meadow grasses, there were no significant differences in soil physical and chemical properties, as compared to a reference site (Armstrong et al 2016). There were some differences in soil biological indicators such as respiration, but the differences were not consistent throughout the year.

Some soil properties may also vary within the solar farm. Unsaturated hydraulic conductivity was found to be higher beneath solar panels on a solar farm in Colorado than at the edge or interspace area between panels (see the schematic of example solar farm in figure 1). The reason for this difference may be the reduced exposure to maintenance activities beneath the panel, which could induce compaction and reduce hydraulic conductivity (Choi et al 2020a). This in turn can impact patterns of soil moisture distribution, which is discussed later in section 3.4.

Zoom In Zoom Out Reset image size

The presence of solar panels has the potential to alter multiple meteorological properties. They may change the balance of incoming solar radiation and emitted radiation, in turn altering soil temperature and evapotranspiration. The fact that they are typically mounted some distance above the ground and inclined may also affect wind dynamics. Additionally, the type and health of vegetation affects actual evapotranspiration (AET) rates, where vegetation may be influenced by the solar panels as well as by human management decisions.

Reduced solar radiation from shading has been documented beneath solar panels on solar farms in France, the United Kingdom, and Oregon and Nevada, USA. This results in lower mean daily soil and air temperatures below the panels in comparison to full sun reference sites during the spring and summer (AL-agele et al 2021, Armstrong et al 2016, Hassanpour Adeh et al 2018, Marrou et al 2013b, Yue et al 2021). Remote sensing of land surface temperature at a solar farm has also indicated overall reductions in soil temperature across the site as compared to pre-development data, though this approach could not explicitly evaluate changes below the panels (Edalat 2017). Wind and humidity changes were not as consistent. Wind speed and vapor pressure deficit did not change relative to solar panels at a solar farm in France (Marrou et al 2013b). However, wind speed, direction, and relative humidity were all altered around solar panels, as compared to the reference site at a solar farm in Oregon (Hassanpour Adeh et al 2018), and vapor pressure deficit decreased under solar panels on solar farms located in the United Kingdom and Arizona USA (Armstrong et al 2016, Barron-Gafford et al 2019).

There can be temporal variability in how solar farms affect meteorological properties and resulting soil dynamics. For example, solar panels decreased the soil temperature beneath the panels compared to reference sites, but these differences were more substantial in spring versus summer (Lambert et al 2021). Increases in soil temperature beneath solar panels relative to reference sites have also been observed during autumn and winter periods at solar farms located in the United Kingdom and western China, when solar panels may help prevent loss of longwave radiation (Armstrong et al 2016, Yue et al 2021). Remote sensing observations of the land surface temperature in Nevada found temperature differences to be greatest in winter when the Sun was lower and the shadows from solar panels were larger (Edalat 2017).

Reductions in solar radiation generally translate into reduced potential evapotranspiration (PET) under solar panels (Elamri et al 2018a, Hassanpour Adeh et al 2018). Solar farm impacts on evapotranspiration dynamics have been investigated in depth on prototypes of agrivoltaic systems in France, testing two crops (cucumber and lettuce) and two solar panel configurations (full panel density vs half density (Marrou et al 2013a). Results show that PET and AET were higher in the full Sun reference locations compared to the solar farm agrivoltaic sites. The reduction in AET was slightly more at the full density solar farm as opposed to the half density site, and differences also varied by crop type. For the lettuce agrivoltaics site, AET over the measured growing period was 103 mm at the full Sun reference, but reduced to 81 and 79 mm for the half density and full density solar farm sites, respectively. Similarly, at the cucumber agrivoltaics site, AET was 178 mm at the full Sun reference, but reduced to 153 and 145 mm at the half density and full density solar farm sites, respectively. In general, ET fluxes were more affected in spring than in summer, indicating temporal variability in the solar panel influence (Marrou et al 2013a). The ratio of transpiration to evaporation also changed, increasing 3 – 4 times in the shaded area of the solar farm sites.

Additional research at this same agrivoltaic site explored the impact of variable panel tilting/tracking on solar radiation and ET (Elamri et al 2018a). While there was still a net reduction in solar radiation and ET under panels with panel tracking, it led to much less heterogeneity than at the sites with a fixed panel orientation.

As with soil properties, vegetation on solar farms is a function of both initial human management decisions regarding the initial solar farm development, and ongoing operations and maintenance decisions on the solar farm. It is also affected by interactions with site soil properties and hydrology. Vegetation is explicitly leveraged as a stormwater management practice, as described later in this review, and is a critical managed element of agrivoltaics operations where solar farms are leveraged for the additional co-benefits of crop production. Thus we feel it is important to review what is known about changes in vegetation on solar farms.

Existing field research has focused on the assessment of vegetation coverage, biomass, and diversity, as well as on interactions with hydrologic processes via water use efficiency or water productivity. Findings are quite variable in these limited studies. Surveys of a solar farm (non-agrivoltaic site) in France largely found no major differences in plant community composition and coverage, relative to nearby reference shrubland and pineland sites. There was a slight increase in the relative abundance of shadow-tolerant plant types under solar panels (Lambert et al 2021). At a reclaimed brownfield site in the Czech Republic, where the land was sown with a meadow grass mixture during solar farm development, differences in plant composition within the solar farm were observed in a survey eight years after the initial development (Uldrijan et al 2021). A greater abundance of taller native perennial grasses was documented in the interspace between panel rows, where more shade-tolerant species and sometimes invasive grasses were observed beneath panels. At a solar farm in the United Kingdom, reduced plant species diversity was observed under solar panels, with reference and panel interspace areas dominated by forbs and legumes (Armstrong et al 2016).

Both increases and decreases in vegetation biomass have been documented under solar panels, depending on the climatic zone. In the areas with lower solar radiation (solar farms in the United Kingdom and the Czech Republic), reduced vegetation coverage and reduced biomass (up to four times lower) have been documented beneath solar panels, relative to panel interspace or reference areas (Armstrong et al 2016, Uldrijan et al 2021). However, the results of research in the solar farm located in Oregon show 126% more dry biomass beneath solar panels relative to the interspace zone and 90% more dry biomass relative to the reference site (Hassanpour Adeh et al 2018). This site is less energy-limited. Thus, the shading of solar panels helps to reduce ET losses and in turn to maximize the water use efficiency of plants, as well as to increase biomass. At crop agrivoltaic sites in Oregon and France, some reduction in crop yield and biomass has been observed under panels relative to nearby reference sites (AL-agele et al 2021, Elamri et al 2018a). At sites with solar panel tracking, biomass was only 16% less than the reference, compared to 30% or greater reductions at fixed panel sites (Elamri et al 2018a).

More efficient water use has been observed under solar panels in multiple cases, mainly because of the shading and reduced solar radiation and PET under the panels. This is particularly evident in locations abundant in solar radiation. In an agrivoltaics solar farm in Arizona, USA, the shade of the solar panels reduced plant drought stress and led to greater crop and food production (Barron-Gafford et al 2019). More efficient water use was also observed for lettuce crops in an agrivoltaics solar farm in France. Coverage of soil by crops was found to be important in reducing soil evaporation and maximizing the availability of water for transpiration and biomass production (Marrou et al 2013a). Water productivity also improved in the shade of panels at a tomato agrivoltaic site located in Oregon (AL-agele et al 2021). Modeling of water and vegetation dynamics at agrivoltaic sites in France successfully reproduced field data on rain and soil water redistribution, and allowed for the additional scenario exploration of plant-water interactions and optimization. The modeled scenarios indicated that the tilting of solar panels could help to minimize water interception and the associated redistribution of water. The scenarios also indicated that agrivoltaics could improve water productivity relative to more traditional agriculture, with only small reductions in crop yield (Elamri et al 2018a, 2018b).

On solar farms, the impervious surface of solar panels intercepts precipitation and drains the water into the interspace between panels. Previously discussed changes in evapotranspiration and soil physical and chemical properties can interact with this altered surface hydrology to cause some heterogeneity in the soil moisture content on solar farms.

Field measurements of the soil moisture on solar farms have often been focused on key locations relative to the solar panels. These have included the interspace area fully open to the sun between panels, under the lower front edge of the panel or 'dripline', underneath the center of the panels, and at the back (higher) edge of the panels where there is partial Sun (figure 1). Sometimes measurements may also be made at a control or reference area outside the main array of panels. At a solar farm in western Oregon, USA on silt clay soils, the soil beneath the center of solar panels was consistently wettest, followed by soil under the back edge of solar panels, and nearby reference soils, with the interspace soils being driest. At a 0.6 m depth, soil under the center of the panels remained near saturation (∼30% volumetric water content (VWC)), whereas the interspace area depleted from ∼30% to ∼20% VWC by the end of the growing season (Hassanpour Adeh et al 2018). Similar patterns have been observed at solar farms in China (Wu et al 2022, Yue et al 2021). Soil moisture at a northwest China site was wettest (10%–20% VWC) at the main dripline at the front of panels as well as under the center of the panel row where a small gap in panels was located; soils at the back edge of panels and nearby reference soils were driest (5%–10% VWC; Wu et al 2022). At an arid western China solar farm, soil moisture was consistently higher under panels—14.7% higher under fixed tilt panels compared to 11% higher under variable tilt panels (Yue et al 2021). Soil moisture was also consistently higher under solar panels at an agrivoltaics site located at Arizona, USA, as compared to an agricultural control site (Barron-Gafford et al 2019). At a solar farm in Colorado, USA (cold semi-arid climate) with paleosols and clay subsoil, some soil moisture variability was observed. Dripline soils were consistently higher (up to 20% – 30% VWC), especially following rain events (Choi et al 2020a). However, substantial variability in soil moisture at all locations relative to solar panels led to a lack of statistically significant differences. The reference site at nearby native grassland was consistently lower in moisture (∼5%). At an agrivoltaic site in France, higher soil moisture was observed at panel driplines, while soil moisture was lower under the panels and in the interspace (Elamri et al 2018b). Another solar farm in France did not demonstrate differences in soil moisture under solar panels compared to the interspace, but did observe overall reduced soil moisture at the solar farm sites as compared to a shrubland reference (Lambert et al 2021). Overall, the main panel dripline at the front of solar panel rows is consistently wetter, and interspace zones tend to be drier. However, under panel soil moisture may vary depending on the balance of evapotranspiration and runoff contributions, due to the climate and panel design.

There have also been some efforts to model soil moisture dynamics on solar farms. These studies have created or leveraged models of rain redistribution from solar panels, and then combined this with various approaches of water and energy balance representation to simulate soil moisture (Elamri et al 2018b, Wu et al 2022). In both cases, there was reasonable agreement between observed and simulated soil moisture. Suspected causes of inaccuracy included challenges in representing the complex energy dynamics under solar panels that influence evapotranspiration (Wu et al 2022) as well as challenges in the accurate representation of water redistribution (Elamri et al 2018b).

To our knowledge, at the time of the writing of this review, the evaluation of runoff generation has occurred only in published modeling-based studies. Many of these studies leverage existing modeling programs, with certain modifications used to represent the unique land cover type of the solar panels. None of these published studies, to our knowledge, have validated their models with field data specific to solar farms.

HEC-HMS (US Army Corps of Engineers 2021) has been leveraged in multiple studies. In one study, the runoff on a solar farm was simulated using a linked model which used a combination of Flo-2D and HEC-HMS (which uses a one-dimensional approach). To simulate the flow from upgradient catchments to the solar farm catchment, HEC-HMS was used. USDA Natural Resource Conservation Service Curve Number methods and shallow water equations were used to predict and route stormwater runoff across FLO-2D grids. The results were reported as maps of maximum flow depths and velocities (Barnard et al 2017a). HEC-HMS has also been used to study hydrologic dynamics in a Nevada, USA (arid climate) solar farm (Edalat 2017). It is shown that regardless of the orientation and tilt angles, runoff volume increases after solar panel installation. Impacts on peak flow are more variable, with the orientation of panels either increasing or decreasing peak flow rates. The results indicate that the panels also noticeably change the rain distribution onto the land surface. Therefore, panel orientation and tilt angles are important factors that need to be considered in stormwater channel design to carry runoff peak flow. One major limitation of this study was that solar panels were represented as an impervious surface on the ground, and infiltration could not be permitted under the panels, as would occur at an actual site; thus this approach likely overestimates runoff (Edalat 2017).

SWAT has also been leveraged for assessing the impact of solar farms on watershed hydrology (Pisinaras et al 2014). For a watershed in northern Greece, scenarios were implemented that induced land use change from agriculture to solar farms in 1 or 5% of the watershed area. Solar farm implementation was represented using soil physical property changes, curve number increase associated with imperviousness, ground cover change from cultivated to bare soil, and reduced solar radiation. The model demonstrated increased surface runoff and percolation, and decreased ET due to solar panel implementation, but these changes were not significant at the watershed scale. However, there is the potential for local-scale impacts. For example, there were increases of ∼100 mm in surface runoff for a given sub-basin, even for a low impact scenario of solar farm implementation (Pisinaras et al 2014).

Other research has relied on custom-built models for representing solar farms (Cook and McCuen 2013). A model written in MATLAB was based on the creation of NRCS type II storms for precipitation (hyetograph) inputs. A simple water balance for each land surface cell was used to allocate precipitation to storage or loss (runoff). Manning's equation was used to estimate runoff velocity and the associated kinetic energy relating to splash erosion. Model scenarios were constructed for a 30 cell grid solar farm, where each cell could have a portion allocated as wet, dry, or interspace. A variety of characteristics were manipulated to simulate changes in soil types, solar panel spacing, vegetation roughness, etc. The results indicated that the addition of solar panels over a grassy field does not change the volume of runoff, the peak discharge, nor time to peak. In general, it was not anticipated that structural stormwater management would be required to prevent adverse impacts (Cook and McCuen 2013).

Some US states, such as Minnesota, have recommended a simple modification to the calculation of impervious surface used in typical runoff modeling approaches for stormwater management planning. Minnesota's recommendation for the modification of runoff calculations leverages the ratio of impervious to pervious surface, where the pervious surface considers both the interspace area as well as the area directly below the panels. Runoff depth associated with this impervious to pervious ratio, as well as soil type, is determined using an Excel tool that leverages the output of an extensive series of models generated in XP-SWMM (Minnesota Pollution Control Agency 2019).

Research on the impacts of solar farms on erosion is quite limited. Some modeling results suggest solar panels can increase erosion. The energy and velocity of water draining from the panels is higher, which could cause erosion in soil below the base of the panels, especially if the interspace is bare. Increases (up to 10 times) of kinetic energy were simulated, which could lead to erosion and the need for erosion control measures, but this modeling effort was not validated by field measurements (Cook and McCuen and 2013).

A larger-scale modeling assessment has quantified changes in erosion and associated sediment loss, along with multiple other ecosystem services for hypothetical solar farms in the Midwestern US. These solar farm scenarios focused on vegetation, comparing hypothetical solar farms with native grassland or turfgrass with baseline agricultural land use. Modeling with InVEST, which leverages the revised universal soil loss equation, estimated sediment export under the solar-native grassland scenario to be 0.007 tons/ha/year, which was a reduction of over 95% and 77% compared to the agriculture and solar-turfgrass scenarios, respectively (Walston et al 2021). However, erosion estimates were based largely on landscape characteristics such as vegetated cover or slope and did not explicitly represent the fact that solar panels are elevated off the ground.

This existing work largely focuses on the potential for erosion on solar farms after initial construction. However, construction of solar farms can require substantial land manipulation. Thus, it is also important to consider this in erosion estimates, and manage this impact appropriately (see discussion in section 4). In addition to erosion associated with runoff, aeolian erosion is a concern in more arid environments, particularly when presence of vegetation is limited (Ravi et al 2016).

There is a small but growing body of work characterizing how solar farm development changes soil properties, vegetation, and hydrologic processes. Most existing work focuses on soil properties, vegetation, soil water, and micrometeorological characteristics or evapotranspiration; there is no published work yet in the academic literature, to our knowledge, documenting the direct measurement of runoff on solar farms.

The results of scientific research vary from findings of no net impact of solar farms on these ecohydrological properties to detection of some impacts. Field soil moisture measurements often demonstrated variability with respect to the solar panels. These changes in soil moisture relative to panels demonstrate the impact of panels on solar radiation, runoff redistribution, and the corresponding evapotranspiration. Particularly in regions where evapotranspiration is not energy-limited (e.g. warm arid regions), there is a climatic sweet spot where the shading of solar panels can help reduce evapotranspiration and maximize water efficiency and facilitate enhanced crop production in association with solar farms (Barron-Gafford et al 2019, Hassanpour Adeh et al 2018). Results from agrivoltaic systems indicate that crops can still achieve high yield under the fluctuating shade of these systems (Marrou et al 2013a). However, more extensive research is merited in agrivoltaic systems across a range of climate settings and crop types.

In wetter climates, there is an interest in keeping soils from reaching sustained levels of saturation and the associated runoff generation. Solar panels introduce heterogeneity in the soil moisture distribution, with precipitation accumulating along the dripline at the lower edge of the panels. It is essential that appropriate management practices are implemented to prevent this heterogeneity from manifesting in increased runoff and erosion generation. Variable panel tilting may also be considered to reduce the redistribution of water.

At the time of this review, we were unable to find any study that directly evaluated runoff generation on solar farms through field measurement. Thus, we are still lacking critical insight into whether solar farms change runoff generation, and whether existing site and stormwater management practices are adequate to prevent adverse impacts. As a result, existing hydrologic models of solar farms are largely uncalibrated. There is also a bias in the sort of sites being evaluated. In general, existing environmental research on solar farms has focused on more ideal sites, i.e. those on sites with lower slopes and well-draining soils. Thus, we are neglecting sites that could be more vulnerable to changes in hydrologic processes with solar farm development.

In general, there is still also a need for simultaneous evaluation of multiple environmental co-benefits from solar farm land management, considering how certain vegetation or crop choices could help manage runoff, but also provide habitat or food.

While there are some environmental conditions that are linked to the inherent characteristics of ground-mounted solar panels (e.g. that there will be at least some level of shading and interception of water from an inclined panel above the ground), many phenomena can be influenced strongly by specific decisions in how the solar farm is constructed and managed.

Changes in certain soil properties (e.g. the reduction in soil organic matter) between solar farms and reference sites indicate the impacts of initial solar farm development, such as some removal of soil and/or vegetation and regrading of soil. This is something that can be changed by a more careful development process. Other soil property changes in different parts of the solar farm (under panel versus interspace) indicate the impacts of continued maintenance between the panels, such as mowing, that may induce some compaction. Thus, these observed detrimental impacts should be addressed as solar farm development guidance is developed. There are opportunities to select management practices that minimize adverse impacts (e.g. soil compaction) and maximize additional benefits- for example, leveraging sheep grazing for vegetation management in lieu of frequent mowing.

Most states in the US currently rely on construction general permits for guiding solar farm development, which may not be adequate for the unique design of solar farms (Great Plains Institute 2021). Only 12 of 50 US states had solar-specific stormwater management guidance, as of the writing of this review. The management practices currently recommended largely leverage low impact development practices of disconnection of the solar panel impervious surface, well-developed shade-tolerant vegetation, and minimal impact of construction practices on soil properties. Where necessary, given site conditions or solar panel configuration, structural stormwater practices such as infiltration practices are recommended. These recommendations are an important start, but validation is needed to confirm that these current practices are adequate, and appropriately tailored to the site conditions. There is a particular lack of guidance relating to appropriate stormwater management practices for solar farms in arid environments (table 3).

Given that runoff volume and quality are key metrics in the stormwater regulatory realm, it is a major gap that there is a lack of field research studying runoff on solar farms. It is also critical that hydrologic modeling approaches be improved to appropriately represent the unique design of solar panels, in that there is an impervious surface with the ability to infiltrate water underneath. In particular, approaches are needed that are simple enough to be widely implemented by those personnel preparing runoff calculations and associated permits for proposed solar farm development.

Though this review focused on hydrology and stormwater management, it is critical that future research and management consider simultaneous evaluation of some of the many other ecosystem services (or disservices) that could be provided by solar farms (Moore-O'Leary et al 2017). The role of solar farms (and how precisely they are developed) can also be explored further within the food-energy-water nexus (Lee et al 2021). Hernandez et al recently proposed a framework of 'techno-ecological synergy' which pushes us to think more broadly about how to implement renewable energy technologies where they will minimize ecological impact and maximize additional co-benefits (Hernandez et al 2019). Thus, instead of simply considering how to minimize the environmental impacts for an already proposed solar farm, we need to also think more broadly about what are the most optimal sites for this type of development. This includes considering how to leverage existing impervious surface for the implementation of utility-scale solar energy, such as parking lots or warehouses. With these multi-criteria and system frameworks, we can ensure that solar farms are developed in the most sustainable manner.