U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA RD-03-081
Date: June 2003
Updated Minimum Retroreflectivity Levels
PDF Version (837 KB)
PDF files can be viewed with the Acrobat® Reader®
In reality, the concept of MR is nearly impossible to specify for each combination of sign type, sign location, driver capability, vehicle type, headlamp type and aim, and viewing geometry that exists on public roads. At a minimum, there can be large differences between sign luminance conspicuity, recognition, and legibility requirements. Regardless of which type of visibility measure one uses, drivers will have different demand levels. For instance, older drivers will generally need higher levels of luminance than younger ones. Vehicles have various dimensions and headlamp performance capabilities that affect nighttime sign visibility. Signs will also have different retroreflective sheeting materials, which redirect light back to the driver at various rates depending on many factors such as the type of retroreflective element, the age of the sheeting, and the orientation with respect to the vehicle and roadway. A number of other issues also can affect the required MR. Therefore, to develop a simple and easy-to-use set of MR recommendations, the research team studied and developed representative levels that were based on the most current information and provided the most reasonable amount of accommodation. It should be noted, however, that no new human performance studies were done during this effort. Rather, the human factors element of updating the MR levels for traffic signs relied on previous literature. Ideally, additional research would be performed to address the assumptions and limitations discussed in this report. A list of recommended research topics is included later.
Sign luminance is the visibility metric that provides the most fundamental groundwork for the establishment of MR levels. Overall, the concept used to establish MR levels results in ML levels needed to read or recognize different signs in different viewing scenarios. Because luminance is the product of the headlamp illuminance and the retroreflective characteristics of the sign material, it is important to have reliable and accurate demand luminance data, headlamp candela profiles, and retroreflective sheeting performance information.
The ML needed to read or recognize a traffic sign is often termed the demand luminance. Demand luminance is dependent on the driver's visual capabilities, the design of the sign legend, and the distance between the driver and the sign. The demand luminance is what researchers try to determine when they study how bright signs need to be before test subjects can read or recognize them.
Illuminance is the amount of energy emitted from the vehicle's headlamp and falling on the signs. Illuminance depends on the headlamp type, the headlamp aim, and the geometry of the particular scenario under consideration.
As mentioned, luminance is the product of the headlamp illuminance and the retroreflectivity of the traffic sign material; this is the supplied luminance. If a particular type of retroreflective sheeting cannot produce a supply luminance at least equal to the demand luminance, it should not be used. If the supply luminance is greater than the demand luminance, then MR levels can be determined. The resulting MR levels represent the point where the demand and supply luminance are equal.
The fundamental concept used to establish MR can be expressed as an equation shown below.
If the Demand RA, NSG is greater than the New RA, NSG, then the material cannot provide the threshold luminance for the given scenario. As shown below, the demand RA, NSG is determined from the illuminance falling on the sign, the viewing geometry (where the Greek letter Nu is the viewing angle for the sign, using the driver as the observation point), and the assumed threshold luminance needed for legibility.
The supply-and-demand concept presented above requires certain assumptions related to both the supply and demand sides of the concept. For instance, on the demand luminance side it is important to have valid luminance threshold data and know what the data represent. Factors such as the subjects' age and the specific research task(s) can have significant impacts on the results of research studies. It is also important to understand the relative distances associated with reported luminance threshold data. Factors that affect the supply luminance side of the overall concept include the viewing geometry, the vehicle headlamps, and the vehicle size, to list a few. The remainder of this chapter discusses the elements of demand-and-supply luminance.
It is important first to recognize that luminance is but one of several measures of sign visibility, and various methods have been used to determine luminance requirements (or demand luminance, as used herein). First, there is the demand luminance measured by, for example, glance legibility, pure legibility, and recognition. Then there are several ways to measure or determine demand luminance criteria. For instance, some researchers have used the orientation of a Landolt C or the letter E; some have used random letters; some have used familiar names; and others have used unfamiliar names. Furthermore, some have used laboratory-based studies with internally illuminated signs or computer monitors; controlled field tests with fixed lighting; or controlled field tests with retroreflected light. There are other important differences as well, but two worth noting are static versus dynamic evaluations and evaluations focused on the general driving population versus the older driving population.
Given these possible bias-generating variations, researchers have wanted to use threshold luminance data derived from actual signs rather than arbitrary targets such as discs used in the Blackwell data, which most visibility models use as their basis. Therefore, the demand luminance work recently completed for white-on-green guide and street-name signs were used at the benchmark to select and study other demand luminance work.(13) The TTI study included 30 drivers aged 55 years or older. The study was performed with full-scale signs in a controlled outdoor environment with retroreflected luminance.
Summed, there were 534 overhead sign observations and 270 street-name sign observations. The results are shown in figure 2 for the legend luminance on a green background (the internal sign contrast ratio during the study was 5:1). Using the results shown in figure 2, ML (or demand luminance) values were generated, as shown in table 2.
These findings were compared to Sivak and Olson's 1985 work related to demand luminance.(32) Their work included geometric means of various luminance studies that had been previously published. They assumed legibility indices (LI) of 50 and 40 feet per inch of letter height for younger and older drivers, respectively. Their recommended demand luminance criteria are shown in table 3 with the LI = 40 ft/in results of the TTI study.
Sivak and Olson's results compare well to the findings presented here. For all three replacement levels, the Sivak and Olson luminance criteria fall between the overhead and street name criteria found as a result of the field studies. It should be noted, however, that the comparison presented here should be interpreted carefully, as the Sivak and Olsen data are based on studies that are not necessarily relevant to today's typical sign design practice, which includes fully retroreflectorized legends on retroreflectorized backgrounds.
Interestingly, for street-name signs, the results of the study are generally higher than for overhead signs or what Sivak and Olson have recommended. One possible explanation for this was that many subjects repeatedly commented on the difficulty they had reading the street-name signs because they perceived that letter spacing was too close. They also commented that the all-uppercase design of the street-name signs made it more difficult to read because of the similarity in word footprints. Had the street-name signs been made with an initial uppercase letter followed by lowercase letters, the threshold luminance values may have been lower.
Unfortunately, the TTI demand luminance data were generated with white-on-green signs and therefore are limited to such signs. Consequently, because of limited time and insufficient funding to support additional human performance research related to sign visibility, the researchers scoured the literature related to demand luminance work. The pertinent literature was critically reviewed to determine how well it compared with the TTI white-on-green data. If strong correlations were found between the demand luminance data found through the literature review and the TTI demand luminance data, then the researchers felt safe generalizing the data for other colors.
For white-on-green signs, the demand luminance work performed by Mercier et al. at FHWA provided a strong correlation to the TTI demand luminance data. Therefore, the Mercier et al. data were used for sign colors other than white on green. A comparison of the TTI and Mercier et al. demand luminance data is shown in figure 3.
Figure 3. Graph. Scatterplot of Data from Mercier et al. and TTI. This graph compares the luminance requirement for white and green signs for people at different ages. On the X-axis is the age of the driver in years from 20 to 80, and on the Y-axis is log of the luminance in candela per meters squared required for the driver to see and read the sign from 0.1 to 100. Three sets of data are being compared: CARTS predicted, Mercer et al., and TTI. The CARTS data shows required luminance levels from 1.5 to 8 as the driver ages, the Mercer et al. data shows required luminance levels from 0.2 to 1.5, and the TTI data shows required luminance levels from 0.3 to 1.8, which are similar to the Mercer et al. data. Comparing these data sets on the graph shows that the CARTS predicted luminance levels are far above what drivers actually need.
To better compare the two data sets, all subjects younger than 55 were excluded from the Mercier et al. data set. Then, using data representing critical detail levels in addition to 1.32 minutes, cumulative distribution plots were generated to compare the spread in the data sets. Figure 4 shows these cumulative distribution curves.
Figure 4 shows that the smaller the critical detail, the more luminance is needed. The TTI data represent Series C legends and the Mercier et al. data represent Series D legends. There is good correlation in this figure as well. Consequently, the researchers felt comfortable using the Mercier et al. data for sign colors other than white on green.
The subsequent analyses of the Mercier et al. data (reported in appendix B) revealed that demand luminance was very low* for certain signs. For instance, for STOP signs, the demand luminance values* were almost always less than 1.0 cd/m2. Therefore, to maintain reasonable levels of conspicuity for iconic signs such as the STOP sign, the demand luminance for all sign types considered was assigned a minimum value of 1.0 cd/m2. This concept is based on comments received at TRB's Visibility Symposium following a project briefing to TRB's Visibility Committee and other participants of the symposium (June 2002, Iowa City, IA). Subsequent discussions among the participants of the symposium, which included many members of the TRB Visibility Committee, revealed that the concept was reasonable in light of the lack of research focused on the subject.
It should be noted, however, that the Mercier et al. data set does not include demand luminance curves for white-on-blue and white-on-brown signs. Additional work is needed to determine appropriate demand luminance data for these colors.
The updated MR levels for traffic signs were derived at distances associated with an LI of 40 feet per inch of letter height (corresponding to the Millennium MUTCD), which results in various distances depending on the assumed letter height. At distances associated with a LI of 40 feet per inch of letter height, Series E letters subtend 1.25 minutes and Series D letters subtend 1.13 minutes of critical detail.
For signs that require maneuvers before reaching the sign (e.g., speed reduction or STOP), the distance provided by an LI of 40 feet per inch is not always valid, especially at higher speeds. In such cases, the distance associated with perception, reaction, and braking time can be greater than the distance provided by using a constant value of LI. For these types of signs (STOP signs), the minimum required visibility distance (MRVD) values from CARTS were used. The MRVD distances comprise a 5-step serial process that includes the distance to:(9)
By using the LI concept, the burden of providing adequate visibility is appropriately placed upon the choice of the proper sign size. The substitution of the CARTS distance introduces the possibility that the sign may be too small for recognition at MRVD. The distance determined by the legibility index is a measure of the critical detail supplied, while the distance determined by CARTS is a measure of the critical detail required. Because signs like the STOP sign often have iconic value, recognition may be accomplished without legibility, so that the critical detail supplied at MRVD is as great as that required. Fortunately, this concept coincides with demand luminance for iconic signs. In the Mercier et al. work, subjects were asked to identify the sign, not necessarily read it. Therefore, for iconic signs such as STOP signs, the demand luminance threshold criteria were likely too low for a driver to actually read the word "stop," but they were high enough to determine that the target was a STOP sign.
Contrast Ratio Issues
The two fundamental elements of traffic signs that allow drivers to understand their intended message are the background and legend, which are designed and manufactured in prescribed color combinations. Drivers rely on the contrast between the background and legend elements of traffic signs to provide legibility. New traffic signs fabricated in accordance with national standards and practices are intended to provide adequate contrast. (3,17,33,34) However, the films and inks used on the face of signs degrade over time, especially when exposed to weather. The degradation rate can vary depending on several factors such as the type of retroreflective film, the color of the film, the compatibility of ink with the retroreflective film, the geographical locale, the direction the sign is facing, and the fabrication techniques. One of the possibly life-ending results of sign degradation is contrast. As the sign degrades, the contrast can reach a point where legibility is unreasonably sacrificed; this level depends on the sign type and the necessary driver actions.
Although contrast can be a problem for a variety of sign types, contrast issues are most relevant to white-on-red signs such as STOP signs, DO NOT ENTER signs, YIELD signs, and WRONG WAY signs. These types of signs are generally made with a process referred to as reverse screening: the sign blank starts with white retroreflective sheeting and then a semi-transparent red film is screened over the white sheeting. The red-screened ink can and usually does fade causing a pinkish to near white appearance. This lightening of the color of red (toward the white region of the color domain) causes the color to become more transparent. Therefore, less of the entering and retroreflected light is absorbed by the color. In other words, the retroreflectivity actually increases (because the sign is able to return more light to the source now that the absorbing dark red color has diminished to a pinkish white color). The end result is that as the sign ages, the contrast between the white legend and red background continues to decrease, eventually approaching a value of one.
A key need related to contrast is the establishment of a criterion that defines the minimum acceptable contrast. The initial set of recommended MR levels for traffic signs required that a contrast of 4:1 be maintained for white-on-red and white-on-green signs. When the recommendations were revised in 1998, the criterion for white-on-green signs was dropped.
The 1993 minimum contrast criterion of 4:1 was chosen based on a review of the literature. Specifically, four sources are referenced in the 1993 report and shown in table 4.
* Findings in foot lamberts (ft-L) were converted to cd/m2 (1 ft-L = 3.426 cd/m2).
The four studies listed in table 4 were conducted to determine the contrast needed to maintain a prescribed threshold of legibility for unknown words or letter orientations. Other studies have also focused on the effect of sign contrast on legibility. For instance, the data plotted in figure 5 are from three studies(35) that tested the orientation of the letter E. The results support the acceptance of a contrast range of at least 3:1, but preferably 4:1 to 50:1.
However, some traffic signs need not necessarily to be read to effectively communicate their message. For instance, STOP signs have a unique octagonal shape and color combination that provides recognition distances much longer than the actual legibility of the legend. Therefore, for such signs, the minimum contrast needed may not be as high as recommended in the initial set of MR levels or in the referenced studies, which focus more on the legibility of a word or orientation of a specific letter.
Recent research conducted at TTI has shown that a STOP sign with a contrast of 3:1 was rated unacceptable by only 4 of 29 State DOT maintenance personnel during a subjective evaluation of 49 different signs on a 5-mile closed-course facility.(26) Although the amount of data is limited, it is not unreasonable to expect that STOP signs and other similar signs with a unique shape and/or color combination require a minimum contrast somewhat less than what is needed to maintain nighttime legibility of unknown words.
Further evidence of the effectiveness of the sign shape and color-coding designs was discussed as early as 1957 when Robinson reported how the National Joint Committee on Uniform Traffic Control Devices (now called the National Committee on Uniform Traffic Control Devices) determined and selected its 12 unique colors.(36) In this article, Robinson demonstrates that color is perhaps the most important element in the code through which traffic control devices convey meaning to drivers. Birren proved how valuable the sign shape and color-coding can be in his work published in 1957.(37) In this study, Birren rearranged the legend of a STOP sign and reported that 86 percent of the subjects passing the sign overlooked the rearranged legend.
It should be noted that other legibility research also supports contrast criteria less than 4:1. For instance, after completing a series of legibility studies in 1976, Forbes et al. concluded that minimum contrast levels of at least 65 percent (2.85:1) are needed to maintain a minimum level of nighttime sign visibility.(38,39) The authors also report that legibility typically levels off at a contrast level of about 80 percent, or 5:1.
The latest version of the British standard for the testing and performance of microprismatic materials includes a range of contrast values depending on sign color combinations.(40) For white-on-red signs, where recognition outweighs the actual legibility of the legend, the minimum is 2.8:1. For white-on-blue and white-on-green, where the predominant visual task is legibility of the legend rather than recognition of the sign, the minimum value is set much higher, at 6.7:1.
There is at least one more reason for using a minimum contrast ratio of 3:1 for white-on-red signs. Based on recent retroreflectivity measurements of Type III (encapsulated) traffic signs in Indiana, Nuber and Bullock present data indicating that the contrast ratio of new white-on-red signs is just above 4:1.(41) Interestingly, their data indicate that the contrast ratio actually increases the longer the signs are weathered. For instance, using the data supplied in their paper, at 2 years the contrast ratio of unwiped white-on-red signs would be approximately 4.7:1. Based on the Nuber and Bullock data, it would not be uncommon to measure a contrast ratio less than 4:1 on new, unweathered white-on-red signs. This is apparently a feature of the screening ink, which has undergone various formulae changes designed to provide more durability in terms of maintaining an acceptable amount of red over time.
One of the only documented and supported arguments against a minimum contrast ratio of 4:1 was presented by Chalmers in 1999.(42) In his report to the Arizona DOT, Chalmers makes a case for lowering the minimum contrast ratio below 4:1, although no recommendation is provided. Using retroreflectivity data from weathering racks in Arizona, Chalmers determined that maintaining a 4:1 contrast ratio for white-on-red signs fabricated with Type I or Type II sheeting would be difficult to do for more than 5 years.
Based on the information described, the contrast ratio can have an effect on demand luminance but the overall effect is small if the ratio is kept within a reasonable range. Provided that signs are made with the same type of retroreflective sheeting or at least logical combinations of retroreflective sheeting (i.e., more efficient materials used for the copy), then the proper color coding is provided and contrast ratio is only an issue for weathering. Therefore, a minimum contrast ratio of 3:1 was chosen as the most appropriate ratio of white-on-red signs (STOP, YIELD, DO NOT ENTER, and WRONG WAY). The designs of these signs produce unique shape and/or color cues that drivers use to recognize the sign before they can read the actual message. The only possible exception is the WRONG WAY sign, but even that is unique in that it is the only red rectangular sign used in the United States. The WRONG WAY sign is also a redundant sign that, by MUTCD standards, is used in conjunction with a DO NOT ENTER sign.
It is important to note that additional research aimed at determining the minimum contrast ratio would be extremely beneficial. This proposed research should be based primarily on the legibility requirements needed for signs with very effective color and shape coding (such as white-on-red signs).
The supply luminance (L) of a retroreflective sign, directed toward the driver, can be estimated as follows:
R A,left and R A,right are the coefficients of retroreflection of the sign corresponding to the vehicle´s left and right headlamps (as source points) with the vehicles driver as the observation point. E left and E right are the separate headlamp illuminance values falling on the sign, measured on planes perpendicular to the respective illumination axis. Nu is the viewing angle for the sign, using the driver as the observation point. Viewing angle (<) and all other retroreflection angles used to determine supply luminance are defined in ASTM E808.(43)
It should be noted, however, that adjustments are needed to account for factors that affect the amount of supply luminance directed from the sign. The luminance calculated from the equation presented above can be thought of as the luminance in a perfect environment with no obstacles between the sign and the observer. However, in a driving environment at least two factors should be considered. The first is the impact of the light scatter caused by the absorption and transmission of light through the windshield. This is called windshield transmissivity and typically reduces the ideal luminance by about 30 percent.(44) The second factor is the atmospheric transmissivity. As light passes through the air, it is scattered by dust particles, and thus the luminance is reduced. Atmospheric reduction factors are available in most physics books and depend on not only the weather conditions but also the viewing distance. An atmospheric transmissivity of 0.53 miles represents clear and dry conditions.(45) Dirty windshields or those that have a haze from cigarette smoke can impede the transmission of luminance even further. (Veiling luminance due to backscatter can also have a significant impact, but is beyond the scope of this report.)
Ascertaining the values in the luminance supply equation involves a combination of photometric and geometric investigation. Most computer models take a systems approach, dividing the data collection into five enterprises, and then a calculation to join them into a single luminance value. The five enterprises are: sheeting photometrics, headlight photometrics, vehicle and driver geometry, sign geometry, and road geometry. The details of these enterprises are presented in a variety of reports and the reader is encouraged to study them for additional detail.(42,46-50)
One way to think of the modeling process is the traditional supply-and-demand concept taught in freshman economics. The combination of the retroreflective sheeting performance data and headlamp performance data along with the associated viewing geometries (defined by the designated sign scenarios) can be used to estimate the supply luminance.
The demand luminance (or the threshold luminance needed by drivers) was based on, among other factors, driver visual capabilities and the required visibility distances. The required visibility distances depend on the designated sign scenario under study. Empirical studies performed in the field and in the laboratory were used to generate threshold demand luminance levels. These studies emphasized the accommodation of older drivers' nighttime needs.
The point of equilibrium between the supply and demand luminance curves can be thought of as the retroreflectivity measuring stick. When the distance associated with the point of equilibrium is less than the required visibility distance, the sign fails to perform at the designated level. When this distance is at least equal to the required visibility distance, a MR level can be generated. However, the viewing geometries of the designated sign scenarios do not correspond to the standard U.S. retroreflectivity measurement geometry of 0.2 and -4.0 degrees for the observation and entrance angles, respectively. Therefore, a conversion was made to change the MR levels at the nonstandard geometries to MR levels at the standard U.S. geometry.
One factor that can affect the MR levels is the viewing geometries associated with the designated sign scenarios. While parameters such as the vehicle dimensions, the vehicle position within the roadway, and the sign position with respect to the vehicle help to define the viewing geometries, one of the more critical parameters of the viewing geometries is the required visibility distance. For the updated MR levels, a modified approach for the required visibility distance was used as compared to the initial set of MR recommendations (published in 1993 and revised in 1998). It is important to note that this modification (i.e., the assignment of the required visibility distance) is only one of several that were implemented as researchers developed updated MR levels for traffic signs.
The updated MR levels were based on required visibility distances associated with a constant legibility index (LI) of 40 feet per inch of letter height (corresponding to the Millennium MUTCD), which results in various distances depending on the letter height of the primary text. At distances associated with a LI of 40 feet per inch of letter height, Series E letters subtend 1.25 minutes and Series D letters subtend 1.13 minutes of critical detail.
For signs that require maneuvers before reaching the sign (e.g., speed reduction or STOP), the required visibility distance provided by an LI of 40 feet per inch is not always valid, especially at higher speeds. In such cases, the distance associated with perception, reaction, and braking time can be greater than the required visibility distance provided by using a constant value of LI. For these types of signs, such as STOP signs, the MRVD values from CARTS were used. The MRVD distances comprise a 5-step serial process involving the distance to:
By using the constant LI concept, the burden of providing adequate visibility is appropriately placed upon the choice of the proper sign size. In other words, the ideal situation would be that the sign size be determined from daytime legibility needs, as determined by the traffic engineer. The nighttime legibility would be maintained as long as the updated MR levels were satisfied. It should be noted, however, that the substitution of the CARTS MRVD distance introduces the possibility that sign size may not be sufficient for recognition at the MRVD. The distance determined by the constant LI is a measure of the critical detail supplied, while the distance determined by CARTS MRVD is a measure of the critical detail required. Because signs (like STOP) often have iconic value, recognition may be accomplished without legibility, so that the critical detail supplied at MRVD is as great as that required and therefore appropriate.