.

Illuminance As A visibility Criterion: The Twilight Envelope (3.2 lux)

Marc Green


The Civil Twilight Method, old science taken out of context. (Zwhalen & Schnell, 1999)

Visibility is determined by physical contrast and contrast thresholds which are subject to a large number of viewer and environment variables. Some have attempted to evaluate visibility and ignore all these considerations by taking the short-cut of using simple illumination measures. One common attempt appeals to the concept of the "twilight envelope".

During civil twilight, approximately the first half hour after sunset, ground illuminance in good weather declines by a factor of about 100, from roughly 330 lux to 3.2 lux. The U.S. Naval Observatory describes civil twilight as a time for "terrestrial objects to be clearly distinguished,"1 an assertion supported by performance data on tasks such as acuity, contrast sensitivity and reaction time. Assuming moderate reflectance, it is also in about the range where many visual functions begin to decline as photopic vision declines and scotopic vision begins.

This has led Owens, Francis and Leibowitz (1989) to propose the visibility concept of "twilight envelope," which suggests that an illumination of 3.2 lux, the level at the end of civil twilight, can act as a rough lower bound for adequate headlamp illumination. Drivers should have reasonably good foveal vision in the area illuminated above 3.2 lux but relatively poor foveal vision when illuminance falls below this twilight envelope. The "twilight distance" is the point down the road where illumination falls below 3.2 lux and the envelope ends. Owens, Francis and Leibowitz (1989) estimated different twilight distances at different heights above the roadway. For headlamps mounted 27 inches high, for example, they estimated distances of 260 feet at the roadway, twilight distances of 175 feet at headlamp level, and only 80 feet at a driver eye height of 42 inches.

The twilight value is an attempt to use illumination as a visibility criterion. All such attempts are unreliable for many reasons. The twilight envelope is very much like the computer PRT program. It tries to take a very complex phenomenon and simplify it to a level where people with no experience or knowledge of the subject matter can claim to know the right answer. Like the use of the computer program, this is wishful thinking:

  • There can be no such thing a single illumination level that is adequate for all tasks. Different tasks require different amounts of light. Reading a newspaper requires a different amount of light from seeing a pedestrian or playing baseball. How could 3.2 lux apply to all situations?;

  • Most importantly, visibility depends on contrast, not illuminance. People do not see light, they see surfaces which are defined by edge contrasts. As explained elsewhere, contrast is the luminance ratio between the pedestrian and the background luminance, as calculated by the Weber fraction. Illumination is only a measure of the light falling on a surface, such as a pedestrian. It also ignores reflectance and hence luminance, which is a critical variable. For example, one study (Wood, Tyrrell, & Carberry, 2005) evaluated detection distances for pedestrians with different clothing under the same conditions. Drivers could respond to pedestrians wearing white clothing at a distance of 52.3 meters and to pedestrians wearing black clothing at only 11.3 meters, a difference of 463%. The twilight envelope concept does not capture such huge effects;

  • The twilight criterion completely ignores the background. In many cases, object contrast may depend on a background that is far in the distance behind the object. For example, the torso of a pedestrian may be seen against a distant background of sky, trees, buildings, etc. The illumination at the pedestrian's location then tells an incomplete story since contrast will depend on illumination (and reflectance) of the distant background. Simply determining illumination (or even luminance) at the pedestrian location says little about contrast and visibility;

  • The twilight value also fails to take account of many other important visual factors such as size, shape, masking, viewing time, visual field location, driver adaptation, lighting uniformity, spectral wavelength distribution, etc. Then there are viewer variables such as age, eye disease, night myopia, the Mandelbaum effect, etc., etc. (see Boff, & Lincoln, 1988, Zwahlen, & Schnell, 1999 and Green et al., 2008 for partial lists of some of the variables). How could a simple illumination criterion predict visibility with such a complicated set of variables?; and

  • No purely photometric measure can specify a detection criterion. There is no such thing as a fixed line between visible and not visible and many cognitive factors, such as expectation affect probability of seeing. These issues have been discussed more fully in Green et al. (2024).

The inadequacy of an illumination criterion for visibility is evident in the development of roadway lighting specifications. They were originally stated in terms of illumination level. This proved so inadequate that specifications switched to more accurate luminance stands. This still proved inadequate for predicting visibility, so the next step was contrast specifications. To make specifications even more accurate, they are now often stated in terms of visibility levels calculated from contrast thresholds. All of this search for the best way to characterize good road lighting has been deemed necessary because simple illuminance values do not accurately predict visibility. Many sources have have noted the problematic nature of any illuminance criterion, e.g.:

As a lighting metric, illuminance is simple to calcu- late and measure, not needing to take into account the reflection properties of the roadway surface and only requiring a fairly inexpensive illuminance meter for field verification. The drawback to this metric is that the amount of luminous flux reaching a surface is often not indicative of how bright a surface will be or how well a person can see. (Lutkevich, McLean, & Cheung, 2012: FHWA Lighting Handbook)


And

Illuminance criteria have been proven to be inad- equate predictors of the effectiveness of lighting systems. Although the visibility of targets is typi- cally directly proportional to illuminance (all other variables held constant), there are too many inter- vening variables that determine the visual stimulus and the efficiency with which that stimulus is pro- cessed by the visual system. (Staplin, Gish, Decina, Lococo, Harkey, Tawneh, Lyles, Mace, & Garvey, 1997)

Conclusion

An illumination criterion is a popular method for specifying lighting requirements because it is simple and easy to measure - you can use an illuminometer (lux meter) which is a relatively cheap light meter rather than a luminance photometer, which is much more expensive.

However, the ease and generality of an illuminance criterion is its vice as well as its virtue. It says nothing about specific situations. It is somewhat analogous to the AASHTO 2.5 seconds perception-reaction time. It is meant to be a normal, worst case number and to apply to the population as a whole, including aged, impaired and distracted drivers. It is not meant to apply to any specific case, where reaction times can be much faster (or sometimes slower). This would be misuse of their PRT value. There is a fundamental difference in the task of AASHTO to find a number that safely covers most of the population as a whole in the general case and the forensic task of determining a reasonable number in a specific situation. An illumination value such as the twilight envelope can be useful, but it is meant only to be a guideline for the general case and should be disregarded when specific facts are known. Even when employed, however, it is at best a very crude and unreliable estimate. It cannot be used to measure contrast and it is contrast that matters.

References

Boff, K. R., & Lincoln, J. E. (1988). Engineering Data Compendium. Human Perception and Performance. Armstrong Aerospace Medical Research Laboratory, Wright Patterson AFB, OH.

Owens D.A., Francis E.L., Leibowitz H.W. (1989). Visibility distance with headlights: A functional approach. SAE Technical Paper Series 890684.

Wood, J. M., Tyrrell, R. A., & Carberry, T. P. (2005). Limitations in drivers' ability to recognize pedestrians at night. Human Factors: The Journal of the Human Factors and Ergonomics Society, 47(3), 644-653.

Zwahlen, H. T., & Schnell, T. (1999). Visual target detection models for civil twilight and night driving conditions. Transportation Research Record: Journal of the Transportation Research Board, 1692(1), 49-65.

Footnotes

1Note that the definition does not say that after the end of civil twilight it is impossible for terrestrial objects to be clearly distinguished.