Supplemental electric lighting for greenhouses may be essential for extending the growing season in northern climates, but it comes with a not-so-hidden cost: environmental light pollution. The consequences of this pollution may range from irate neighbours in rural areas to municipal bylaws that may prohibit the use of supplemental lighting during certain hours of the night or altogether.
Searching for information on light pollution online will yield plenty of results. However, it soon becomes apparent that it does not apply to greenhouse lighting. Most of the discussion concerns astronomical light pollution, which is of interest to both amateur and professional astronomers, and where the focus is on outdoor lighting for roadways and parking lots. There are also discussions of “light trespass” onto residential properties from adjacent street lighting, which is rarely a concern for commercial greenhouses.
In addition, there are discussions of the effects of “artificial light at night” (ALAN) on nocturnal wildlife, including insects, fish, amphibians, and mammals. This is a complicated topic, as the effects differ between species and genera. There is almost nothing, however, that is specific to supplemental lighting for commercial greenhouses.
This can be a problem, particularly if citizen action committees lobby government agencies for bylaws and regulations. While alleviating light pollution is a laudable goal, it is necessary for all sides – concerned citizens, municipal authorities, and commercial greenhouse operators – to know the facts and discuss the matter accordingly. Beyond this, there needs to be agreement on what can reasonably be done to alleviate any problems.
At present, the primary source of information for citizen groups are publications from the International Dark-Sky Association (www.darksky.org) in North America and allied organizations in Europe (e.g., www.savethenight.eu). In Canada, the Royal Astronomical Society of Canada has its Light Pollution Abatement Program (www.rasc.ca/lpa) with similar goals. The goal of this article is therefore to look at light pollution in the context of greenhouse supplemental lighting.
Astronomical Light Pollution
Concerns over light pollution began in the 1950s, when the light from large cities began to interfere with astronomical observations made from nearby observatories. (The Dominion Astrophysical Observatory, for example, is only 10 km away from downtown Victoria, BC.) A combination of urban and suburban growth, coupled with the changes from incandescent street lamps to more efficient (and much brighter) mercury vapour and later high-pressure sodium (HPS) lamps, continually exacerbated the problem.
The problem is that even on a clear night, light emitted by roadway and parking lot light fixtures (aka “luminaires”) is reflected from the ground into the night sky. While most of this light escapes into outer space (as we can clearly see when flying over cities at night), a small but significant portion is backscattered down towards the ground by air molecules and airborne dust and smog. The resultant “sky glow” obscures our view of the fainter stars (particularly the iconic Milky Way), and of course interferes with observations by both professional and amateur astronomers.
By itself, a single roadway luminaire does not contribute significantly to sky glow; it is the cumulative effect of thousands to tens of thousands of luminaires in a city that is the problem. Large commercial operations, however, may have thousands of horticultural luminaires and so be equivalent to a small city in terms of light pollution. (Hybrid greenhouses for cannabis production, for example, may have as many as eight hundred 1,000-watt HPS luminaires per acre.)
In rural areas away from city smog, the relative amount of backscattered light from roadway luminaires is dependent on the spectral power distribution (or more colloquially, “spectrum”) of the light source. In particular, air molecules (primary nitrogen and oxygen) scatter blue light more effectively than red light (which is why the sky appears blue to us).
Figure 1 shows the spectra of typical high-pressure sodium (HPS) and light-emitting diode (LED) lights used for greenhouse supplemental lighting. The HPS luminaires look yellow-orange to us, while the LED luminaires appear, depending on the ratio of red to blue light (between 1:1 for vegetative growth to 6:1 for reproductive growth), as various shades of purple (or “blurple”). Figure 1 assumes a 1:1 ratio, which is the worst-case scenario for light pollution. The spectra have been scaled such that both light sources produce equal amounts of photosynthetically active radiation (PAR).
Following IDA and similar publications on astronomical light pollution, concerned citizens and municipal engineers may say something like this: “The amount of backscatter increases as the fourth power of the inverse of wavelength. The blue light of horticultural LEDs will produce much more light pollution than HPS!” This is correct, but it does not explain the situation very well. Figure 2 illustrates what this mean: the amount of backscattered blue LED light is roughly twice that of the backscattered HPS light.
This however does not tell the full story. When our eyes are fully dark-adapted on a starry night, we are most sensitive to blue-green light (actually 505 nm) and less sensitive to yellow and blue. When we take this into account, the increase in astronomical light pollution when changing from HPS to 1:1 blue-red LED lighting is a walloping 3.8 times.
Ecological Light Pollution
Less often talked about but perhaps even more important are the effects of electric lighting on animal and plant life. It is impossible to cover all of these effects here, in part because ecological light pollution is the subject of intense ongoing research by biologists and ecologists. Rodents such as mice are most sensitive to green light and ultraviolet radiation, some migratory birds and bats are most affected by green light, insects respond to ultraviolet radiation, and plants can be impacted by red and far-red light, which disrupts their growth and development. Mammals, reptiles, amphibians, birds, insects … perhaps the best that can be said is that any excess light at night can be considered ecological light pollution, regardless of its spectrum.
While little research has been devoted to the topic to date, there is also the question of whether light pollution from greenhouses (FIG. 3 and FIG. 4) adversely impact birds, bats, and insects. In urban areas, tens of million migratory birds die every year from nighttime collisions with office tower windows. The US Fish and Wildlife Service is more specific – an estimated 6.6 million migratory birds die each year from collisions with communication towers and their warning lights. While migratory animals may not die from collisions with greenhouses, they are likely disoriented by the bright lights and may suffer increased mortality rates due to exhaustion.
Concerned citizens may lobby their municipal councils to “take back the night” so that their children can experience the starry nights they may remember as children, but little attention is paid to overcast nights, when the light pollution can be much worse.
On a clear night, the amount of backscattered light is miniscule – we only notice it because the night sky without light pollution can be exceedingly dark. Low-level stratus clouds, in the other hand, can reflect up to 75 percent of the light reflected from the ground.
For greenhouses, it all depends on the cloud ceiling height (FIG. 4). If it is 2,000 meters, any light from the greenhouse facility reflected from the clouds will be spread over an area of perhaps ten square kilometers and be reasonably unobtrusive. However, if the ceiling height is 250 meters, the area within a radius about 1500 meters will receive up to 60 times the amount of light pollution on the ground. If the greenhouses are near a sensitive ecological area, this can be a problem.
The amount of light pollution on the ground also depends on the average ground reflectance. For most of the year, this is between 5 and 10 percent. During the winter months with snow on the ground, however, it can increase to anywhere from 40 to 90 percent. Figure 5 shows the differences for summer (solid lines) and winter (dashed lines) ground conditions assuming 10 percent and 80 percent ground reflectance respectively.
The visibility of light reflected from the clouds is another matter entirely, as this depends on the cloud opacity and whether the clouds can be seen on the horizon on an otherwise clear night. Brightly illuminated clouds may disorient migratory animals, again possibly increasing mortality rates due to exhaustion. They may also lad to complaints from residents living a considerable distance from the greenhouses.
The simplest solution for most light pollution situations is to use the minimum amount of light needed, and turn the luminaires on only when needed. Unfortunately, this is rarely practical for greenhouse operations. Each crop has specific PAR and daily light integral (DLI) requirements. In response to the suggestion that fewer luminaires could be used or the luminaires turned on for fewer hours at night, each crop has specific requirements related to its circadian rhythms. Like animals, plants need to sleep at night. There may be some flexibility in the supplemental lighting schedule, but likely not enough to make a difference.
What will make a difference are blackout curtains on both the walls and roof of the greenhouse that can be closed at night. The name “blackout” notwithstanding, the preferred color facing inwards is white. In a large greenhouse where the floor area is much greater than the wall area, most of the light from the luminaires will be reflected from the plant canopy and ground through the roof panels. If the average floor reflectance is 10 percent and the curtain reflectance is 80 percent, the plant canopy will receive an additional 20 percent of light.
If a white polyethylene ground cover is used for the greenhouse floor, the average floor reflectance (assuming walkways between the plants) will likely be in the range of 40 to 50 percent. The light will in this case bounce back and forth between the floor and ceiling at least ten times before it is essentially absorbed by the plants, curtains, and floor cover. Each bounce provides further PAR for the plant canopy, so that the end result may be over 100 percent of additional light.
Whatever the amount of additional light received by the plant canopy, it is quite literally recycled light. It reduces the need for supplemental lighting, and can be seen as an operating cost offset to the capital and operating cost of the blackout curtains (which can also function as energy curtains in cold climates).
It is also worth noting that this recycled light is diffuse, and is directed at the plant leaves from both above and below. For many greenhouse crops, this can be beneficial, resulting in stronger stems and less leaf senescence.
In communities with large concentrations of greenhouses, such as Leamington, ON, light pollution is receiving increasing scrutiny from the press. You know you have a problem when the Detroit Press runs a major article on the light pollution from greenhouses located 25 miles away from the city.
LTO Nederland, the Dutch greenhouse industry organization, has been proactive in mandating screening for 98 percent of the greenhouse if lighting is being used during nighttime hours. If the light levels are greater than 15,000 lux during the hours of 5:00 PM to midnight, the screens need to block 98 percent of the light during the entire night, with the side screens blocking a minimum of 95 percent during nighttime hours.
If greenhouse operators wait until their municipalities begin discussing regulations and bylaws, it may put them at a considerable disadvantage. With so little information available to municipal engineers and planners to formulate proposed regulations, it may be advisable to be proactive.
There is no question that supplemental electric lighting in greenhouses can cause significant environmental light pollution. Having facts and figures available when discussing the issue with concerned citizens and municipal authorities is useful; having a solution that may reduce operating costs is a bonus.
In these unfortunate times, it seems that everyone is looking for ways to help with the COVID-19 crisis. From a lighting designer’s perspective, one solution is obvious: ultraviolet radiation. We have known about the disinfectant properties of ultraviolet radiation for nearly 150 years (Downes and Blunt 1877), and we have been using low-pressure mercury arc lamps to kill bacteria and inactivate viruses for the past 85 years (Wells and Fair 1935).
My initial goal in writing this article had been to discuss ultraviolet germicidal irradiation (UVGI) methods and materials, but this work was pre-empted by the Illuminating Engineering Society’s Photobiology Committee and their excellent report, IES CR-2-20-V1, Germicidal Ultraviolet (GUV) – Frequently Asked Questions.
I can highly recommend this publication for any questions related to ultraviolet disinfection and germicidal lamps, as it is written so as to be accessible to both lighting professionals and the general public. However, there is one small but critical omission that needs to be addressed.
Many consumers are currently interested in ultraviolet sterilizers for home use. Earlier this month, Forbes published an article titled, “Seven of the Best UV Sterilizers for Phones and Other Household Objects.” These devices included a USB UV sanitizer wand, two UV smartphone cleaning cases, a UV water bottle purifier cap, a portable UV sterilizer for small items such as baby pacifiers, an ultraviolet LED sanitization box, and a 25-watt germicidal lamp. Six of these are reasonable and perfectly safe if used correctly. The seventh … well, we need some background to explain the problem.
The ABCs of Ultraviolet
The Commission Internationale de l’Eclairage (CIE) divides the ultraviolet region of the electromagnetic spectrum into three subregions according to wavelength:
In the natural environment, the only significant source of ultraviolet radiation is sunlight. As the solar spectrum shown in Figure 1 shows, this radiation is only a small fraction of the total radiation received from the Sun. Our atmosphere fortunately absorbs much of the carcinogenic UV-B radiation at sea level, and essentially all of the UV-C radiation, thanks to the Earth’s ozone layer. The risk of exposure to UV-C radiation is therefore entirely from electric light sources.
Now, there are several types of light sources that emit UV-C radiation, but only low-pressure mercury arc lamps and UV-C light-emitting diodes (LEDs) are found in consumer products. We can therefore ignore high-pressure mercury arc lamps, pulsed xenon arc lamps, and krypton-chlorine excimer lamps – this article is about ultraviolet disinfection solution for consumers.
In 1931, F. L. Gates demonstrated that UV-C radiation effectively kills bacteria and protozoa, and inactivates viruses. Using a high-pressure mercury-vapor arc lamp, he produced an “action spectrum” showing effectiveness versus wavelength (Gates 1931). He further noted that this spectrum was very similar to the absorbance spectrum of deoxynucleic acid (DNA), and correctly surmised that the radiation was terminally disrupting the DNA molecules (Figure 2).
The advantage of low-pressure mercury arc lamps is that they emit monochromatic UV-C radiation at 254 nanometers, which is close to the 265 nm peak of Gates’ action spectrum. Even better, these lamps are basically the same as compact and linear fluorescent lamps, but with two important differences: 1) they do not have a phosphor coating to convert the ultraviolet radiation into visible light; and 2) they use fused quartz or soda-lime “soft” glass rather than borosilicate glass for their envelopes. (Fused quartz is transparent at UV-C wavelengths, whereas borosilicate glass is opaque.)
UV-C Exposure Limits
The problem with germicidal lamps is that exposure to UV-C radiation can irritate the eyes and skin. These irritations are technically photokeratitis (“snow blindness” or “welder’s flash”), photoconjunctivitis (“pink eye”), and erythema (sunburn). This medical terminology belies the seriousness of the issue, but a short paper by Trevisan et al. (2006) puts it into perspective.
In 2005, twenty-six medical students attended a 90-minute anatomy lesson in an autopsy room at the University of Padova in Italy. There were three autopsy tables, above each of which was mounted a single 40-watt germicidal lamp. A timer turned the lamps on at 7:00 PM and off at 7:00 AM; the lesson began at 8:30 AM.
Unfortunately for the students involved, the timer failed. They were dressed in protection suits against biological risk, but their eyes and the skin of their faces, scalps and necks were exposed to UV-C radiation for the duration of the lesson. As is common with UV-C exposure, the onset of symptoms occurred on average four hours afterwards. For the eyes, these included a burning sensation, excessive tearing, pain, blurry vision, hindrance to open, conjunctival redness, eyelid swelling, and photophobia. For the skin, the symptoms included acute sunburn, a burning sensation, irritation, and pain. The paper further noted an ocular “foreign body sensation,” which translates into a feeling of having sand constantly in your eyes. These problems lasted two to four days, but six students afterwards reported persistent scurf (flaking of the skin), and one of persistent dryness of the skin.
There are two important points here: 1) the lamps were basically 40-watt linear fluorescent lamps; and 2) the exposure time was only 90 minutes. The horizontal irradiance was later measured and found to vary between 50 μW/cm2 at the level of the autopsy table to 250 μW/cm2 one meter below the germicidal lamps. To put this into perspective, a horizontal illuminance of 500 lux is equivalent to approximately 250 μW/cm2 of visible light.
The American Conference of Governmental Industrial Hygienists recommends a maximum UV-C exposure dose limit for a workday of 3.0 millijoules per square centimeter (ACGIH 2020). For a period of 90 minutes, this translates into an effective irradiance of 0.6 μW/cm2. The students exceeded this recommendation by a factor of 100 to 400 times.
UV-C at Home
You might dismiss the misfortune of the medical students as an unfortunate workplace accident. After all, who would want to have a high-power germicidal lamp in their home?
… which brings us to the seventh UV sterilizer listed in the Forbes article:
The description of this lamp reads, “UV Germicidal Bulb 25w E26/E27 Screw Socket UV Light Bulb 110 Volt, UVC Ozone Free.” It was priced at $35 on 2020/04/02, but is currently (two weeks later) priced at $75. It is clearly a product that is in demand, although the description is less than reassuring:
“When this bulb lit, will immediately have a special smell in the air, it means it worked, this special smell comes from burnt harmful smalls by the UVC ray, just like the smell in the hospital.”
Th manufacturer explicitly states that this product has a quartz envelope and emits 253.7 nm UV-C radiation, but … 25 watts? Considering the medical distress that the medical students suffered from 40-watt lamps, surely this is incorrect?
Apparently not, for the manufacturer also claims that it will sanitize up to 400 square feet (37 square meters) in 15 to 60 minutes, making it “suitable for basement, bathroom, school, canteen, kindergarten, home, salon, closet, hotel … shoe cabinet, dog/chicken house, toilet …” and so on. The Forbes reviewer helpfully added that, “This UV bulb can be screwed into any standard light socket in your home or office.”
Nowhere on the Amazon Web page is there any mention as to how dangerous the output from a 25-watt germicidal lamp can be. “Wait,” you say, “surely this product has to comply with the relevant government regulations designed to protect consumers from such unseen dangers?”
If you are an American, the Occupational Safety and Health Administration has this to say as a standards interpretation: “… there are no OSHA-mandated employee exposure limits to ultraviolet radiation.” If you are a Canadian, the Canadian Center for Occupational Health and Safety says, “There are no Canadian regulatory occupational exposure limits for UV radiation. Many jurisdictions follow the limits recommended by the American Conference of Governmental Industrial Hygienists (ACGIH).”
There are government standards regulating the emission of UV-B radiation from suntanning beds and ultraviolet emissions from white-light mercury vapor lamps … but apparently nothing for UV-C disinfectant lamps. In other words, dear consumer, caveat emptor.
With no regulations to constrain them, manufacturers of high-power UV-C disinfection lamps are free to sell whatever they want to the unsuspecting consumer … and Amazon is assuredly there to help them. Enter “150 watt UV ozone” in the search bar for www.amazon.com and you will get over 140 results like this:
There are a number of reasons to be suspicious about the manufacturer’s claims for this product, but suppose we take the 150-watt electrical input power at face value. We can reasonably assume that the lamp ballast (remember that it is essentially a linear fluorescent lamp) has an efficiency of 90 percent. This means that 135 watts is being delivered to the mercury-vapor arc.
The radiant efficiency of a low-pressure mercury arc is about 45 percent, while the UV-C transmittance of fused quartz is about 90 percent, and so the lamp is emitting 55 watts of UV-C radiation.
The lamp itself, despite its outsized appearance in FIG. 4, is about 80 cm high. If we were to place a 60 cm diameter by 80 cm long tube over this lamp, it would have an inside surface area of roughly 15,000 cm2. The UV-C irradiance of this surface would therefore be 5.5 x 107 microwatts divided by 1.5 x 104 square centimeters, or about 3,650 μW/cm2 – over 6,000 times the ACGIH-recommended limit for 90 minutes’ exposure!
How can you tell whether the unit is activated? Ultraviolet radiation is by definition invisible, but someone sitting next to the unit in the dark, possibly watching television, may eventually notice a pretty violet glow around the lamp electrodes (Fig. 5). This will alert them that they will become mightily uncomfortable two to six hours later.
One reason to be suspicious about the manufacturer’s claims for this product is that it supposedly generates ozone. According to the manufacturer, “Ozone has sterilization and disinfection, in addition to formaldehyde and odor, ozone can fill the entire room without being affected by obstructions.”
Whatever the claim of formaldehyde refers to, ozone is a particularly toxic gas, even in minute concentrations of 0.1 parts per million (ppm). It smells like chlorine gas, and it can cause similar (and severe) damage to respiratory tissues. A low-pressure mercury arc emits both 254 nm and 185 nm UV-C radiation. The shorter wavelength radiation is capable of ionizing the molecular oxygen in the air surrounding the lamp and so create ozone, but the quartz or soft glass envelopes of UV disinfection lamps are doped such that they are opaque to 185 nm radiation. Some UV-C lamps do emit ozone, but they are only used for industrial applications such as municipal water disinfection (where it replaces chlorine).
Product safety information? The description for a similar product reads, “Do not look directly at the germicidal light source. Absorbing too much ultraviolet light can cause skin irritation and conjunctival damage.” This is regrettably what we get in the absence of enforced government regulations.
There are reasons to suspect some of the manufacturers’ descriptions of these products (many of which are identical apart from the manufacturer’s name), particularly when one manufacturer states that the product has an “ozone mode” and an “ozone-free mode,” where the latter is suitable for pregnant women and the elderly. How the doping of a quartz envelope designed to block 185 nm radiation from ionizing the surrounding air can be toggled on and off with the simple flip of a switch is left unexplained.
With over 140 products like these to choose from, the average consumer should feel … absolutely petrified.
But rest assured! If you are concerned about your health, searching for “UV disinfection COVID” yields at least 20 UV sterilizer products that are available for purchase. One blatantly states, “Kill 99% COVID-19,” while another features a 10-watt battery-powered flashlight marketed as a “UVC Germicidal Lamp” with “Brightness: about 500 lumens.”
My favorite, however, must be the “UV COVID Indoor Sterilizer” shown in Figure 6. Lighting designers will understandably suspect that this is a theatrical LED luminaire, but the manufacturer assures us that it (somehow) emits “Ultraviolet wavelength of 253.7 nm.” Oh yes, and it can also be used for aromatherapy!
We are now living in different times, but one thing will always be the same: wherever there is a dire situation, there will be those among us willing and able to take advantage of our fears and gullibility. Say the words “ultraviolet disinfection” today and you will appreciate what technology has brought us – an online marketplace for charlatans and their wonderous pills that cure all ills.
For Europeans, the relevant standard concerning ultraviolet radiation in the workplace is Table 1.1 of EU-OSHA Directive 2006/25/EC, which specifies a maximum daily exposure to UV-C radiation of 30 J/m2, which is equivalent to the ACGIH recommendation of 3 mJ/cm2. The related publication “Non-binding Guide to Good Practice for Implementing Directive 2006/25/EC, Artificial Optical Radiation” (EC 2010) is highly recommended as an information source.
Worldwide, ISO 15858:2016, UV-C Devices – Safety Information – Permissible Human Exposure applies. This standard is applicable to in-duct UV-C systems, upper-air in room UV-C systems, portable in-room disinfection UV-C devices, and any other UV-C devices that may cause UV-C exposure to humans.
I am indebted to my colleague Dieter Lang of LEDVANCE for the following information and illustrations from his LinkedIn posting, “The Visible Spectrum of Mercury Low Pressure Lamps for Disinfection”:
He also provided a useful and informative visible light spectrum for these lamps:
As Dieter writes, “The color shown in the photograph above is quite close to the real visual impression … of course you need to avoid looking at such a lamp without suitable eye protection.”
With this, I readily acknowledge that Figure 5 is somewhat misleading — it should be evident that high-power UV-C germicidal lamps are activated when viewed in a darkened room. Unfortunately, it was impossible to find photographs of such lamps in use that had not been doctored for advertising purposes (e.g., FIG. 4) by the manufacturer.
ACGIH. 2020. Ultraviolet Radiation: TLV(R) Physical Agents 7th Edition Documentation. Cincinnati, OH: American Conference of Governmental Industrial Hygienists.
Downes, A., and T. P. Blunt. 1877. “Researches on the Effect of Light Upon Bacteria and Other Organisms,” Proc. Royal Society of London 26:488-500.
EC. 2006. Directive 2006/25/EC – Artificial Optical Radiation. European Commission.
EC. 2010. Non-binding Guide to Good Practice for Implementing Directive 2006/25/EC, Artificial Optical Radiation.” European Commission.
Gates, F. L. 1931. “The Absorption of Ultraviolet Light by Bacteria,” J. General Physiology 14:31-42.
ISO. 2016. ISO 15858:2016, UV-C Devices – Safety Information – Permissible Human Exposure.
Trevisan, A., et al. 2006. “Unusual High Exposure to Ultraviolet-C Radiation,” Photochemistry and Photobiology 82:1077-1079.
SCHEER. 2017. Scientific Committee on Health, Environmental and Emerging Risks: Opinion on Biological Effects of UV-C Radiation Relevant to Health with Particular Reference to UV-C lamps, European Commission.
Wells, W. F., and M. G. Fair. 1935. “Viability of B. coli Exposed to Ultraviolet Radiation in Air,” Science 82:280-281.
Ian Ashdown, P. Eng, FIES, Senior Scientist, SunTracker Technologies Ltd.
Whether you call it “circadian lighting,” “biologically effective lighting” or some other name, the principle is the same: the color and intensity of light can be used to regulate the timing of our biological clocks, or “circadian rhythms.” For architects and lighting designers, this is an opportunity to provide healthy and comfortable environments for building occupants.
From an academic perspective, circadian lighting represents the culmination of over two decades of research into the effects of light on circadian rhythms. While there remain a number of open questions and ongoing research to address them, it has been argued that we now know enough to translate this knowledge into practice with building code standards and recommended practices for architectural lighting design. From an engineering perspective … not so fast. A look at three metrics shows why.
WELL Building Standard
The WELL Building Standard v2 with Q1 Addenda is dedicated to the concept of building designs that promote healthy environments for living, working, learning and play. One of its hundreds of design guidelines is Feature 54, Circadian Lighting Design (https://standard. wellcertified.com/light/circadian-lighting-design).
The underlying concept is simple: predict or measure the Equivalent Melanopic Lux (EML) incident on the vertical plane at the eye level of the occupant. For work areas, the design requirements are then:
At 75% or more of workstations, provide at least 200 EML (including daylight if present) at four ft above the floor facing forward (to simulate the view of the occupant) between the hours of 9:00 am and 1:00 pm for every day of the year; or
For all workstations, provide maintained illuminance of at least 150 EML on the vertical plane facing forward.
There are similar requirements for living environments, breakrooms and learning areas. Unfortunately, architectural lighting design programs such as Lighting Analysts’ AGi32 and DIAL’s DIAL Evo do not predict EML. They do, however, predict photopic vertical illuminance, EV. All the designer has to do then is to calculate or measure EV values and multiply them by the melanopic ratio, R.
How do you calculate this ratio? These are the instructions from Table L2 of Feature 54: “To calculate the melanopic ratio of light, start by obtaining the light output of the lamp at each 5 nm increment, either from [the] manufacturer or by using a spectroradiometer. Then, multiply the output by the melanopic and visual curves given below to get the melanopic and visual responses. Finally, divide the total melanopic response by the total visual response and multiply the quotient by 1.218.”
The International WELL Building Institute helpfully provides a downloadable Excel spreadsheet to perform these calculations, which includes six sample spectra for common light sources—easy. Once again, from an engineering perspective, however … not so fast.
Questions, Questions …
Questions regarding the WELL Building Standard, or at least Feature 54, arise when considering how architectural lighting design is performed in practice. For example:
Unlike measurements of horizontal illuminance (EH), vertical illuminance (EV) measurements require both a specified position and a direction for the meter sensor. Four ft above the floor to “simulate the view of the occupant” makes sense, but it overlooks the reality that information on workstation locations and their associated furniture is often unavailable during the design phase.
Obtaining the spectral power distributions (SPDs) of luminaires from the manufacturer is in most cases all but impossible. This may change in the future as lighting design and analysis software programs become capable of utilizing this information directly, but for now it is mostly necessary to either manually digitize the manufacturers’ printed datasheets (if available) or measure representative products with a spectroradiometer. This is rarely practical during the design phase, when it is unlikely that specific products will have been identified.
Handheld spectroradiometers for field measurements are readily available, but they typically have a spectral resolution of 8 to 10 nm. They may report spectral power distributions in 5-nm or even 1-nm increments, but these are interpreted values. Depending on the light source (including particular fluorescent lamps and LED modules with “spiky” distributions), the calculation of equivalent melanopic lux from EV measurements may be insufficiently accurate. Following CIE recommendations for spectral metric calculations, a spectral resolution of at least 5 nm is required.
For workstations, what about the computer display monitors that the occupants will presumably be facing for most of the workday? With luminance values in the range of 250 to 350 candelas per sq meter, display monitors provide considerably more vertical illumination than the surrounding cubicle walls do, and they further have SPDs similar to white light LEDs of 6500K. If anything, they likely contribute more to the circadian lighting as perceived by the workstation occupants than the room lighting does. If they are to be considered, how should they be modeled? More important, what is their angular subtense in the occupant’s field of view? A dual 27-in. display fills much more of the occupant’s field of view than does a 17-in. display, for example.
Should daylight be included in the predictions or measurements of vertical illuminance? The amount of daylight entering an interior work space depends on the time and date, the building orientation and windows configuration, the sky condition (clear to overcast), the glazing transmittance, whether the blinds are open or closed, and the office partitions and furniture layout. Moreover, what is the SPD of the daylight on clear and overcast days? (The CCT of daylight can vary from 5500K for direct sunlight to more than 25000K for clear blue poleward sky.) Also, in modeling daylight, this should include the daylight diffusely reflected from the exterior ground, as it typically comprises some 10-20% of the daylight entering the space on overcast days.
How should multiple light sources, each with their own SPDs, be modeled? The occupant’s position and orientation at a “typical” workstation might include, depending on the time of day, direct sunlight, diffuse daylight, direct illumination from overhead luminaires, and indirect light from possibly strongly colored surfaces that are illuminated by both overhead luminaires and a desk lamp. It is all but impossible to predict the contributions of these light sources to the vertical illuminance, let alone calculate the resultant composite SPD as seen by the observer.
To put these questions into context, imagine the WELL design requirements being incorporated into Section 16500 clauses of a building specification contract document. It is one thing for an engineer or lighting designer to follow the WELL requirements as guidelines and feel comfortable in saying that the design is compliant; it is quite another when a contractual dispute arises or a building inspector decides to take in situ measurements. From a legal perspective, such ambiguities in the specifications are definitely not a good thing.
UL is currently preparing UL 24480, Recommended Practice and Design for Promoting Circadian Entrainment with Light for Day-Active People, a document that is based on the Circadian Stimulus (CS) metric developed by Rensselaer Polytechnic Institute’s Lighting Research Center. One of the key features of this document is that it relies on the user accessing the LRC’s CS Calculator (lrc.rpi.edu/cscalculator/), a web-based tool for converting predicted or measured vertical illuminance values into CS metric values.
Unlike the WELL spreadsheet, the CS Calculator offers close to 200 different lamp SPDs for the designer to choose from. If, for example, you are interested in 3500K LED light sources, there are 19 SPDs to choose from, while for 4000K LEDs there are 20, and eight for 5000K … and therein lies a problem. For the designer, how do you choose?
Like the WELL’s EML metric, the CS metric is calculated from predicted or measured vertical illuminances, with minimum recommended CS values for different situations. If, for example, the design requires a CS value of 0.30, the designer chooses a light source from the list of options (or provides a custom SPD with 5-nm resolution); the calculator then determines the vertical illuminance needed to achieve this value.
The problem is that even with the same correlated color temperature, the range of required vertical illuminances for the different LEDs is shocking (Figure 1). To be fair, some of these LED light sources are unconventional, including violet-pump LEDs (approximately 415 nm) with triphosphor coatings, and hybrid white-light LEDs combined with deep-red LEDs to boost the CRI R9 values. However, there is nothing to stop the designer from randomly choosing one of these products as a “typical” LED light source for calculation purposes.
If nothing else, Figure 1 makes one point perfectly clear: there is no reasonable relationship between CCT and the CS metric, at least for 3500K and 4000K LEDs. The range of vertical illuminances for 3500K LEDs is 3:1, while that for 4000K LEDs is 2:1. Randomly choosing an LED product as representative of all LEDs with the same or similar CCT could lead to problematic consequences.
Compared to the 3500K and 4000K LEDs, the range of vertical illuminances for 5000K LEDs is quite small—only +/- 9%. The reason for this is that the eight LEDs appear to have almost identical SPDs above 500 nm, which is entirely due to the yellow-emitting phosphor blend. The only significant differences are the peak wavelengths of the blue-pump LEDs, which vary from 440 to 450 nm. This situation could, however, change with future developments in LED and phosphor technologies.
Additionally, what if the building specification permits “or equal” substitutions for luminaires? The CS metric value depends on the absolute spectral irradiance incident on the observer’s corneae, and so the luminaire product specification sheet would have to provide a graph of EV versus CS. Further, depending on the EV tolerance the designer is willing to accept, many more products may be disqualified compared to evaluation on luminous flux alone. As any experienced commercial electrical engineer will attest, this can lead to rather heated discussions between the engineering and architectural firms or with the electrical contractor.
Furthermore, the CS Calculator allows the designer to specify multiple light sources and combines their SPDs into a composite SPD as seen by the occupant. However, it is left as an exercise for the designer to calculate the relative contribution of each light source—including indirect light from possibly colored surfaces in the office space or wherever (and not to mention daylight with its range of possible CCTs)—to any predicted vertical illuminance. It is possible to do this with existing architectural lighting design programs, but only if the designer is willing to model and calculate the spatial light distribution in the environment separately for each type of light source. In addition, these programs do not take light source SPDs into account apart from their CCTs, so the results would be at best approximate.
EML versus CS
This is for the CS metric — what about the EML metric? Fortunately, the CS Calculator provides relative SPDs for each of its light sources, with resolution of 2 nm. Rescaling these SPDs to 5-nm resolution and using the WELL spreadsheet produces the melanopic ratios (R) shown in Figure 2.
There is still a 60% variation in the EML values for a given vertical illuminance value for LEDs of 4000K. This may be preferable (from an engineering perspective) to the 3:1 variation for the CS metric, but it is still an unreasonably wide range for lighting design purposes. It also begs the question: what are the criteria for choosing either the WELL or CS metrics apart from tolerances? Which circadian lighting metric better represents the effect of the lighting on circadian rhythm entrainment? Simply choosing the metric that offers the best results borders on unethical engineering practices.
In summary, we may now “know enough” about the effects of light on our circadian rhythms to design circadian lighting. Whether this is true is debatable, but it is also beside the point. From an engineering perspective, it is abundantly clear that we do not have the calculation tools needed to predict or measure circadian lighting metrics, including EML and CS, to acceptable engineering standards.
For reference, it is generally accepted that architectural lighting design software is capable of predicting horizontal and vertical illuminance values to within +/-10%, given reasonably accurate surface colors and reflectance values. With circadian lighting metrics, however, we are confounded by the variations between different LED products with the same CCT.
We are further confounded by what exactly it is we are expected to predict with our lighting design software or measure in the field. Simply saying “workstations” and “vertical illuminance” belies the complexity of both our architectural environments and our behaviors in them. This is not something that any standard or recommended practice can ever hope to reasonably address—there are simply too many variables for the designer to consider.
Regardless of what we
may know about the effects of circadian lighting on human health and wellbeing,
we may never be able to codify this knowledge in building design practices. It
is reasonable for standards organizations to offer design guidelines based on
what we know, and it is highly recommended that lighting designers learn the
principles and benefits of circadian lighting. These should not, however, be
codified as inflexible design requirements.
Ian Ashdown, P. Eng., FIES, Senior Scientist, SunTracker Technologies Ltd.
Look at a greenhouse manufacturer’s product specifications and you will see that the light transmittance of single-pane clear glass is typically 88 to 91 percent. Compared to double-wall polycarbonate with a transmittance of 80 percent, it would seem that glass is the better choice. However, if you measure the photosynthetically active radiation (PAR) at the leaf canopy within the greenhouse, it is often only 40 to 60 percent of that measured outside the greenhouse. Why is this?
The answer is that these transmittance values were based on standard test procedures developed by the American Society of Testing Materials (ASTM), which require the incident light to be perpendicular to the glazing material. For greenhouses however, the incident light comes from direct sunlight, diffuse daylight, and daylight reflected from the ground and other exterior surfaces. In other words, light is incident upon the glazing material from all angles.
To better understand the issue, look at a sheet of clear glass. When it is perpendicular to your line of sight, it is essentially transparent. However, as you tilt the glass, you begin to notice reflections. These reflections increase in brightness until you are looking at the glass at a grazing angle, at which point it behaves essentially like a mirror.
This also happens, of course, with daylight entering the greenhouse. On a clear day, this is mostly direct sunlight, and so the amount of sunlight entering the greenhouse depends on the incidence angle θ (Fig. 1).
Assuming that the glazing material is perfectly transparent (that is, it does not absorb any light), the PAR light transmittance varies with the incidence angle as shown in FIG. 2.
To put this into perspective, consider a gutter-connected greenhouse with a 1:2 (30-degree) roof pitch and double-pane glass glazing that is located in Vancouver, Canada at a latitude of 49 degrees and oriented on an east-west axis. The solar elevation at noon on December 21st will be 18 degrees. The incidence angle will be 42 degrees, and so the transmittance of the south-facing roof panels will be 48 percent. We can now see where the “40 to 60 percent” figure come from.
It is important to note that these results apply only to clear glazing materials with smooth surfaces, such as glass and acrylic. For materials such as polyethylene and polycarbonate with rough or striated surfaces that tend to diffuse the incident light, it becomes more difficult to predict their optical characteristics. A better approach is to measure their transmittance for various incidence angles in a test laboratory.
Predicting the precise amount of daylight that will be incident upon the
leaf canopy in a greenhouse can be done, but it requires a horticultural
lighting design program that considers building latitude and orientation,
building layout and dimensions, glazing materials, date and time, weather
conditions, and more. For now, however, it is sufficient to see why the
measured light at the leaf canopy is considerably less than what is measured
Ian Ashdown, P. Eng., FIES, Senior Scientist, SunTracker Technologies Ltd.
Most LED grow lights feature blue and red LEDs whose peak wavelengths – approximately 450 nm for blue and 660 nm for red – have been chosen to coincide with the spectral absorption peaks of chlorophyll A and B molecules. In doing so, they optimize the conversion of electrical energy into plant photosynthesis.
Some manufacturers, however, are now offering grow lights with “far-red” LEDs that feature peak wavelengths of approximately 735 nm. Unfortunately, they offer little if any information on why these LEDs are useful.
In order to make an informed choice when purchasing these grow lights, it is necessary to understand some of the science behind far-red radiation and how plants perceive and respond to it.
Red and Far-Red Radiation
What we call “visible light” is electromagnetic radiation with wavelengths ranging from 400 to 700 nanometers (nm). We perceive this radiation as ranging from very deep blue (400 nm), bordering on ultraviolet radiation, to very deep red (700 nm), bordering on infrared radiation. Coincidentally, this is also the range of wavelengths that plants can utilize for photosynthesis (PAR = Photosynthetically Active Radiation).
There is no formal definition of “red” in terms of wavelength, but it is often considered to consist of wavelengths ranging from 600 nm (bordering on orange) to 700 nm. The term “far-red,” on the other hand, has been formally defined to consist of wavelengths between 700 nm and 800 nm. We can barely see this radiation as a very deep red if the radiation is intense enough, but it is for practical purposes invisible to the human eye. Plants, on the other hand, readily perceive and respond to far-red radiation.
We see vegetation as being green because the chlorophyll A and B molecules strongly absorb blue and red light. A typical green leaf absorbs 90 percent of incident red light; the remainder is reflected and transmitted (FIG. 1). Beyond 700 nm, however, chlorophyll is basically transparent. This means that beyond approximately 750 nm, green vegetation reflects 40 percent and transmits 55 percent of far-red radiation. The region of rapid change in spectral reflectance between 700 and 750 nm is called the “red edge,” and is used to monitor vegetation coverage from space using remote imaging.
Angiosperms – flowering plants – also take advantage of the red edge using a family of photoreceptor molecules called phytochromes (Latin for “plant color”). They have been doing so since they first appeared in the fossil record some 160 million years ago. It is not an exaggeration to say that without these molecules, we would still be living in a world of conifers, cycads, and ginkgoes.
The number of different phytochromes varies by plant species – rice has three, thale cress (Arabidopsis thalania) has five, maize has six, and so on. Each type serves different (and often multiple) functions in each species, but they all absorb red and far-red light in exactly the same manner.
Each phytochrome molecule has two states called isoforms. Left in the dark for several hours, it reverts to a state called Pr, where it strongly absorbs red light (FIG. 2). If a phytochrome molecule in this state absorbs a red photon, it changes to its Pfr state, where it absorbs far-red radiation. If the molecule absorbs a far-red photon, it reverts back to its Pr state. When in its Pfr state, the molecule is biologically active, and may interact with the plant’s molecular machinery. Given this, phytochrome can be seen as a reversible biological switch that can enable or inhibit various plant functions. One such important function is the detection of neighboring plants.
Flowering plants that tolerate full or partial sun need to gain access for their leaves to direct sunlight in order to photosynthesize. The problem is that they often have competition from other plants for the same resource. That is, the leaves of other plants may block access, either at present or in the future. In response, the plant may elongate its stem (apical dominance) and decrease branching in order to tower above the competition. In doing so, it necessarily diverts resources from other priorities, including producing secondary metabolites for pathogen resistance and insect herbivore deterrence, improving drought tolerance, and reducing root biomass. Together, these responses are called the shade-avoidance syndrome (SAS).
If the plant realizes or predicts that it cannot avoid being shaded, it responds by growing in a more compact form and flowering early. Being crowded by other plants, it is more susceptible to pathogen and herbivore predation. The best strategy is therefore to build chemical and defenses and stay close to the ground while producing seeds as soon as possible in order to survive into the next generation.
Shade-loving flowering plants, on the other hand, may not exhibit any of the SAS responses, at least to the same degree. The daylight they receive has likely been diffused by the forest canopy, and so there is less advantage in devoting resources to avoid being blocked by the leaves of neighboring plants.
On a clear day, direct sunlight has a ratio of red light to far-red radiation (R:FR) of about 1.3. That is, there is about 30 percent more red light than far-red radiation that is received by the plant leaves. Even daylight reflected from natural inorganic materials such as rock and soil exhibits roughly the same R:FR ratio.
When the direct sunlight is being blocked by the leaves of neighboring plants, however, the “red edge” effect takes hold. A single layer of leaves can change the R:FR ratio from 1.3 to 0.2 or less. That is, there is now about six times less red light than far-red radiation incident on the plant leaves. Two layers of leaves and the difference becomes thirty times or more.
Flowering plants use phytochrome to detect the R:FR ratio and so decide whether SAS responses are necessary. In addition to detecting whether the direct sunlight is being directly blocked, the plants can determine from the R:FR ratio whether there are neighboring plants that might pose a future threat and so initiate appropriate SAS responses.
End of Day
The R:FR ratio of direct sunlight is about 1.3 during most of the day, but it approaches 0.6 or so during twilight when the atmosphere preferentially scatters blue light and the sky turns yellow and red. This only lasts for half an hour or less, but it is important because plants use these changes to synchronize their internal circadian clocks both with the 24-hour day and the seasons. This involves a burst of gene expression activity that is controlled by phytochrome.
Blackout curtains can be used in greenhouses to eliminate twilight, and both red and far-red LEDs can be used to simulate twilight for vertical farms at the end of the daily photoperiod. Interestingly, low PAR values are required, on the order of one µmol·m-2·sec-1, for this purpose. Various SAS responses to red and far-red pulses have been recorded for different species, including stem elongation and changes in leaf area. End-of-day pulses of far-red radiation, for example, have been shown to result in useful hypocotyl elongation of tomato rootstocks for grafting.
Floriculturists have long used incandescent lighting at night to disrupt the photoperiod of short-day plants such as poinsettias and chrysanthemums. During the night, the phytochrome molecules revert to their biologically inactive Pr state. If the plants are exposed to incandescent lighting (which has an R:FR ratio of 0.7) during the night, the phytochrome molecules are re-activated, which results in their circadian clocks being reset. Repeated nightly exposure (“night breaks”) in the middle of the night prevents the plants from stopping vegetative growth and setting their flower buds.
For long-day plants, night break lighting may have the opposite effect of advancing rather than delaying flowering. The operative here is “may,” as different species and even cultivars respond differently to night breaks.
It is important to note that only red light can be used for night breaks; when phytochrome is in its Pr state, it cannot absorb far-red photons. Red LEDs with their 660 nm peak wavelengths are thus ideal for night break lighting, whereas far-red LEDs will have no effect (FIG. 2).
Far-Red Radiation Sources
Given that plants are subjected to an R:FR ratio of 1.3 in direct sunlight and much lower ratios when shaded by neighbouring plants, it is interesting to consider what we subject them to with various electric light sources. Incandescent lamps have an R:FR ratio of 0.7, which is what plants would perceive when they are adjacent to neighboring plants but not directly shaded.
High-pressure sodium (HPS) lamps, on the other hand, have an R:FR ratio of about 4.8, metal halide lamps have R:FR ratios varying from 2.6 to 3.4. and white light LEDs (regardless of color temperature) have R:FR ratios varying from 3.6 to 4.0. Various fluorescent lamps have R:FR ratios varying from 5.5 to 13.0 and above.
The common reason for these high R:FR ratios is that, putting aside technology limitations, the lamps are designed for visual applications – there is no reason for them to generate invisible far-red radiation. If they did, it would simply lower their luminous efficacy (lumens per electrical watt) values.
With only blue and red LEDs, the R:FR ratio is essentially infinite. SAS responses can sometimes be elicited by blue light alone, but the likelihood is that many plants will not recognize the presence of neighboring competitors when irradiated by most grow lights.
This leaves open many questions regarding the possible applications of far-red LEDs. While various species and cultivars may grow well (or not) under the familiar blue and red (“blurple”) LEDs, they may not exhibit any shade avoidance syndrome responses. In some situations, these responses may actually be desirable. For instances, sun-loving plants that are grown in shade may be more compact, but they may also exhibit greater pathogen and drought resistance, and they may generate desirable secondary metabolites as defense mechanisms. Their flowering may be advanced or delayed, the number of buds may change … the list goes on.
It is also possible that end-of-day far-red pulses of radiation at low irradiance levels may have a greater effect on plant growth than during the day, as this is when gene expression is particularly active. (Blue light pulses at dawn have also been shown to change plant morphology.) One advantage is that this requires less energy than having the far-red LEDs continuously on during the day.
For horticulturalists and floriculturists, experimentation with far-red LEDs offers opportunities for developing species- and cultivar-specific light recipes as trade secrets. If horticultural luminaire manufacturers have not yet said why far-red LEDs are useful, it is because there is much that still needs to be researched and discovered. With a basic knowledge of the science behind far-red radiation and the phytochromes, it becomes practical to experiment with light recipes and photoperiods, and to understand why the plants respond the way they do.