IMSA Journal Feature Article
May/Jun 2003
IMSA Journal

John D. Bullough, Lighting Research Center, Rensselaer Polytechnic Institute
Published In LD+A, December 2002

What do we need traffic signals for?
Most North Americans love to drive, and one of the more ubiquitous forms of lighting on our roads are traffic signals. Among the advantages of these lighting devices, according to the Millennium Edition of the U.S. Manual on Uniform Traffic Control Devices (MUTCD, online at are orderly traffic flow, increased traffic capacity, and reduced frequency of crashes. Arguably, the reduction of crashes is paramount, because crashes can result in the loss of human life. For this reason, the MUTCD refers to important standards about the color and the required luminous intensities of traffic signals, so that they can be detected quickly and accurately as we drive.

What do the standards say?
In North America, these standards1,2 give the minimum luminous intensities required for 8- and 12-inch signals of each of the three colors (red, green and yellow). Most interestingly, green signals are required to have a luminous intensity that is 2.0 times that of the red signal, while yellow signals require 4.6 times the luminous intensity of the red signal. These differences diverge from practice in Europe and Japan, where all three signal colors are required to have the same luminous intensity.

These differences lead to certain questions: Do yellow and green traffic signals in North America need to be brighter than the red signal? In particular, do yellow traffic signals, which have nearly five times the luminous intensity of red signals, need to be as bright as they are? This last question is particularly relevant because of the advent of traffic signals using light-emitting diodes (LEDs) instead of incandescent lamps. Red LED signals are becoming almost commonplace throughout North America, with green LED signals following closely behind. Generally, transportation agencies consider the installation of red and green LEDs to be very successful (they use 80 percent less energy than their incandescent counterparts, and are warranted to last much longer). However, no standards-complying yellow LED traffic signal currently exists on the market. Meeting the higher luminous intensity standard with yellow LEDs results in a product that uses too many LEDs to be cost-effective, and in problems with heat dissipation and maintaining light output.

Why does yellow matter?
One could argue that the lack of a standards-complying yellow LED signal is unimportant. Many agencies are converting to LED traffic signals because of energy and maintenance savings, but the yellow signal is only on for a few seconds during each signal cycle, which means yellow LEDs would not significantly affect the overall life-cycle cost. This is no doubt the case for many retrofit situations, but it is also true that many intersections convert to a flashing mode late at night, where the yellow signal is flashed continuously in one direction and the red signal in another. Here, potential energy and maintenance savings from yellow LED signals could be meaningful. They might become even more meaningful in those instances where battery backup is provided in case the electricity supply is interrupted; using LEDs in flashing mode at these times might give work crews a few extra hours to handle potential emergency situations!

What is the basis for the higher luminous intensity of yellow (and green) signals, anyway? Unfortunately there doesn't appear to be any readily available documentation, but a hint is found when we consider the relative transmittance of the colored glasses used in incandescent signals. Green glass lets about twice as much light through as red glass; yellow glass four to five times as much. And since the red signal is arguably the most critical in terms of life safety, it might have been perceived by early standards makers as most important to "get red right" and then to scale the luminous intensities of the other two colors by the increase in transmittance afforded by green and yellow glass. (Of course, this remains pure speculation.)

What does the research say?
The Lighting Research Center at Rensselaer Polytechnic Institute recently completed research that can help standards makers grappling with differences among various international specifications.3 Against a bright background that simulated daytime viewing, subjects were asked to detect simulated incandescent and LED traffic signals of varying luminous intensities and colors. If the signal was not detected within one second, it was considered a missed signal. In this way, the reaction times and percentage of missed signals could be gathered. Thousands of trials were collected systematically in this study. Of interest, there were no differences in visual response between incandescent and LED signals having the same color and luminous intensity. A red LED signal performed just as well as a red incandescent signal. There were, however, significant differences among the three signal colors. For the same intensity, red signals consistently resulted in shorter reaction times and fewer missed signals than yellow and green signals.

These findings suggest that for traffic signals in Europe and Japan, which are specified with equal luminous intensities, yellow and green signals result in longer reaction times and in a greater percentage of misses than red signals. Is such an approach appropriate? Certainly, there does not appear to be any controversy abroad associated with traffic signals of different colors having the same luminous intensity. Still, one could argue that cultural differences among drivers in North America, Europe and Japan make differences in traffic signal intensities desirable in North America. Again, looking at the results from the aforementioned study can shed some additional light.

As described above, red LED traffic signals meeting the North American standards have been widely used in what seems to be a successful manner. Since these signals seem to be successful, their resulting reaction times and percentage of missed signals are probably acceptable in practice. We also know that yellow and green signals having the same luminous intensity as red signals will result in longer reaction times and a greater percentage of misses, which might be considered unacceptable in North America (although this is clearly not unacceptable in Europe and Japan). Using the results of the Lighting Research Center's study, one can determine how much higher yellow and green signals must be in order to have the same reaction time and the same percentage of misses as a red, standards-complying signal.

Using an equivalent reaction time as the criterion, the luminous intensity multiplication factors are 2.0 for the yellow signal and 2.6 for the green signal. Using an equivalent percentage of missed signals as the criterion, the luminous intensity multiplication factors are 1.4 for the yellow signal and 2.8 for the green signal. In other words a yellow signal that is twice the intensity of a standards-compliant red signal will result in a reaction time and a percentage of missed signals no worse than the complying red signal. If, and only if, this is a reasonable criterion for the performance of a yellow traffic signal, then the existing standards that require yellow signals to have a luminous intensity that is 4.6 times that of the red signal are clearly much higher than needed. As an aside, reducing the luminous intensity of the yellow signal will also help to make them less uncomfortable to view at night. Yellow and green signals meeting current standards can often be considered uncomfortably bright; red signals are almost always not.4

What about the green signal? The results above indicate that a green signal would need to be nearly three times higher in luminous intensity than the red signal to perform as well as a complying red signal in terms of reaction time and missed signals. This is a pretty large increase over what is currently required by standards. Should standards therefore increase the luminous intensity required of green signals? It seems unlikely. The type of response that is required of a green signal (start, or continue moving) is very different than that required of the red and yellow signals (stop!). There do not appear to be problems with current green traffic signals in North America, which are "only" twice the luminous intensity of red signals. It is therefore unlikely that requiring green signals to have the same reaction time and percentage of missed signals as a red signal, is a reasonable requirement. Furthermore, increasing the luminous intensity of green signals would make them more uncomfortable at night than they already appear to be.

Nonetheless, one could argue that it is also an unreasonable criterion for the yellow signal, too. Perhaps it is even more important to respond quickly and accurately to the yellow signal than to the red signal, since after all, it is a precursor to the red signal. If this is the case, the data from the study described above can assist in identifying just how much higher the luminous intensity should be. Other data too will be important in answering fundamental questions about appropriate luminous intensities, especially for drivers with color deficiencies, and older drivers. The results from the studies described here, as well as from previous studies and from a soon-to-be-published study commissioned by the National Cooperative Highway Research Program, will be especially helpful as requirements for traffic signals are revisited.

What next?
It could be argued that even considering a reduction in the luminous intensity of traffic signals is irresponsible. After all, any reduction in luminous intensity will result in longer reaction times and a greater percentage of missed signals. Whether this reduction in visual performance is meaningful, is a more difficult question to answer. In addition, concerns about energy use and costs of maintenance nationwide have prompted some agencies to begin using non-complying yellow LED signals, regardless of the lack of standards-complying yellow LED products. These issues make it ever more important that standards makers consider what visual responses are required for each signal color and how a given luminous intensity provides the required visual response. Without careful, yet timely, consideration, the examples of "going-it-alone" that have already begun will grow while standards-making bodies will grow less relevant. Such precedents will surely affect organizations like the IESNA, which develops standards and recommendations for many applications. Standards makers should act proactively now rather than reacting later to these inevitable questions.

1. Institute of Transportation Engineers. Vehicle Traffic Control Signal Heads: A Standard of the Institute of Transportation Engineers. ITE, Washington, DC, 1985.
2. Institute of Transportation Engineers. Vehicle Traffic Control Signal Heads - Part 2: Light Emitting Diode Traffic Control Signal Modules. ITE, Washington, DC, 1998.
3. Bullough, J. D., P. R. Boyce, A. Bierman, K. M. Conway, K. Huang, C. P. O'Rourke, C. M. Hunter and A. Nakata. Response to simulated traffic signals using light-emitting diode and incandescent sources. Transportation Research Record 1724. TRB, Washington, DC, 2000.
4. Bullough, J. D. P. R. Boyce, A. Bierman, C. M. Hunter, K. M. Conway, A. Nakata and M. G. Figueiro. Traffic signal luminance and visual discomfort at night. 80th Annual Transportation Research Board Meeting. TRB, Washington, DC, 2001.

IMSA Journal Feature Article
May/Jun 2003
IMSA Journal