The Influences of Bright–Dark Lighting Environments on Driving Safety in the Diverging Zone of Interchange in Highway Tunnels

Abstract

Increasing the lighting luminance in the diverging zone of interchange in highway tunnels can generally enhance driving safety. However, it creates a bright–dark luminance contrast with the adjacent road. A pronounced contrast can induce new driving risks. This underlying mechanism remains unclear. Three key factors, i.e., the luminance of the dark environment, the bright–dark luminance ratio, and the position of the small target, are identified in this paper, which affect drivers’ visual recognition abilities. Based on fundamental tunnel lighting design rules, a series of naturalistic driving tests on the visual recognition distance for small targets with 132 conditions were designed. It combined three dark environment luminance levels (1.5~3.5 cd/m²), four bright–dark luminance ratios (2~5), and eleven small target positions (−50~+50 m). Twenty-four drivers were randomly selected and drove vehicles under the different scenarios. Their visual recognition distances for small targets were recorded and analyzed. The results show that visual recognition distances for small target visuals under different bright–dark lighting environments vary significantly, and the shortest distances occur exactly at the luminance boundary. Both decreasing the bright–dark luminance ratio and proportionally increasing the luminance levels of the bright and dark environments can markedly improve the visual recognition distance. A multi-parameter regression model was developed to correlate the visual recognition distance at the bright–dark luminance boundary with the luminance of the dark environment and the bright–dark luminance ratio. Based on drivers’ required safe sight distance, a method for setting lighting luminance in the diverging zone of interchange was proposed. The methodology and findings offer technical support for lighting design and safety management in the diverging zone of interchange in highway tunnels.

1. Introduction

With the continuous expansion of urban underground roads and highway tunnels in both construction scale and operational mileage, the number of underground interchange projects facilitating traffic transitions within underground road networks or between underground and surface roads has grown significantly [1]. Among them, the diverging zone of underground interchanges have become high risk accident areas due to concentrated traffic conflicts caused by frequent lane-changing maneuvers among vehicles [2]. Creating a reasonable lighting environment to meet drivers’ visual task requirements is crucial for the safety of traffic in diverging zones. Goswamy et al. found that after installing lighting at intersections, nighttime accidents decreased by 33% based on analysis of nighttime accident data [3]. Scott analyzed the relationship between average road luminance and the nighttime-to-daytime accident ratio, concluding that when road luminance increased from 0.5 cd/m² to 1.0 cd/m², the nighttime-to-daytime accident ratio dropped from 57% to 42% [4]. Both domestically and internationally, engineering measures that increase illumination in the diverging zone have been adopted [5]. This approach aims to enhance drivers’ visual perception [6] and vigilance [7], thereby improving overall traffic safety. The International Commission on Illumination (CIE) states in its research report that the luminance level at interchange merging/diverging zones should exceed that of the approaching roadways [8]. China’s Guidelines for Design of Lighting of Highway Tunnels stipulates that the luminance in tunnel diverging zones should not be less than three times that of the basic interior zones [9]. This differentiated lighting design results in a marked luminance disparity between the interchange diverging zones inside the tunnel and the adjacent interior zones, giving rise to two distinct lighting environments characterized by a bright–dark contrast, as shown in Figure 1. When a driver travels through such tunnel, moving from the bright environment into the dark environment, the excessive luminance contrast can create a “black-hole effect” in his visual field. This impairs their ability to discern the road ahead, often prompting an abrupt deceleration that may in turn trigger new traffic conflicts and safety hazards [10]. Therefore, investigating the bright–dark lighting environmental characteristics of tunnel interchange diverging zones that preserve drivers’ visual ability is essential for improving road safety.

Current studies on how bright–dark lighting environments influence driver’s visual perception and traffic safety have predominantly examined the daylight in the tunnel entrance scenario. Regarding research methodologies and evaluation metrics for the bright–dark lighting environment quality, Schreuder et al. firstly recreated various combinations of exterior luminance, tunnel threshold zone luminance, and small target luminance in the laboratory. They used whether subjects could detect the small target within a set time as the criterion to investigate how the relationship between the threshold zone and exterior luminance influences the driver’s visual ability [11]. Narisada et al. employed a similar experimental methodology, progressively scaling the tunnel entrance and target dimensions in proportion to distance and speed during the experiment, thereby investigating the test results and evolving patterns under dynamic conditions [12]. He et al. conducted visual-performance experiments by simulating a threshold zone lighting environment and measuring and analyzing drivers’ reaction times to detect a small target so as to evaluate the lighting quality of the threshold zone [13]. Due to the differences between the simulated lighting environment in the laboratory and the real tunnel threshold zone lighting environment, the validation of the corresponding research conclusions in engineering applications is limited [14]. Some researchers have carried out investigations directly at tunnel entrances, but it is difficult to collect data such as the reaction time and detection probability during actual vehicle tests. To investigate the impact of tunnel entrance lighting environments on drivers’ visual cognition capability, researchers conducted naturalistic driving recognition tests using small targets at a tunnel entrance based on the recommendation by CIE. They measured drivers’ visual recognition distances to these targets when approaching tunnels at various speeds from outside of the entrances [15]. Based on the driving visual task, i.e., requiring timely detection of road obstacles within a safe distance, this approach evaluates whether the tunnel threshold lighting meets safety requirements [16]. The achieved conclusions better align with actual driving needs of safety.

In the studies of the factors affecting drivers’ visual recognition ability in the bright–dark lighting environment at a tunnel entrance, researchers have explored and analyzed three aspects: luminance reduction factor at the threshold zone, color temperature of the lighting source, and the target’s position. Among them, the luminance reduction factor at the threshold zone is defined as the ratio of the threshold zone of luminance inside the tunnel to the luminance outside the tunnel. A higher factor means the lighting luminance at the threshold zone more closely matches the exterior luminance. Based on collected visual recognition distances of drivers detecting small targets at the road surface, Hu et al. established a positive correlation between the visual recognition distances and luminance reduction factor at the threshold zone. They concluded that higher factors correspond to increased visual recognition distances and enhanced visual capabilities for small target detection [15]. The color temperature of lighting sources in tunnel threshold zones also affects drivers’ visual recognition distance. Ma investigated the impact of lighting source color temperatures on drivers’ visual capability in tunnel entrances The study revealed that when the natural light color temperature is below 6000 K, a higher lighting source color temperature in the threshold zone correlates with longer driver visual recognition distances. Conversely, when the natural light color temperature exceeds 6000 K, a lower lighting source color temperature results in extended visual recognition distances [17]. Besides the impact of lighting characteristics on drivers’ visual recognition distance in tunnel entrance zones, Wang and Zhang also analyzed the relationship between the location of road obstacles in tunnel entrance zones and drivers’ visual recognition distance. They found that the visual recognition distance of obstacles in the tunnel entrance zone shows a V-shape variation trend with its location, and the shortest visual recognition distance occurs when the obstacle is located approximately 60 to 80 m away from the tunnel entrance [14,18].

By investigating the effects of various factors on drivers’ visual recognition distance in the bright–dark lighting environment at tunnel entrance zones, researchers have proposed different luminance reduction factors at the threshold zone to optimize the luminance of the tunnel threshold zone. The range of luminance reduction factors at the threshold zone obtained in the relevant research findings is between 0.01 and 0.07 (with luminance ratios of approximately 100 to 14) [9,19]. Due to the significantly lower overall luminance value of the bright–dark lighting environment in tunnel interchange diverging zones compared with that at tunnel entrances, maintaining equivalent visual recognition distance capabilities for drivers requires distinct luminance ratios in different scenarios [11,12]. The luminance ratio of the bright–dark lighting environment at the tunnel entrance zone is not applicable to the diverging zone of the tunnel interchange. Otherwise, it is likely to form the “black-hole effect”, which poses traffic safety risks. At present, in the research on the lighting environment of the diverging zone of the tunnel interchanges, Ding et al. have analyzed the various patterns of lighting quality characteristics such as road surface illuminance and uniformity by changing parameters like the layout, spacing, and mounting angles. However, there is still a gap in the research on how the luminance ratio of the bright–dark lighting environment in the diverging zone of the tunnel interchanges affects drivers’ recognition abilities [20].

Based on an analysis of factors affecting drivers’ visual recognition capabilities in bright–dark lighting environments, this study aims to design a naturalistic driving recognition test plan involving multivariate-coupled tunnel lighting conditions for small targets. Field tests are conducted to collect drivers’ visual recognition distance data for these targets. The research further analyzes how bright–dark lighting environments in tunnels influence the visual recognition distance and driving safety, which provides a reference basis for luminance design in the diverging zone of tunnel interchanges.

2. Mechanisms and Influencing Factors

In order to ensure the driving safety of drivers, the quality of road lighting environments should be able to ensure that drivers’ visual recognition distance of obstacles on the road consistently exceeds the safety distance required to take necessary actions [21]. Meanwhile, the stopping sight distance is recognized as the minimum safety distance threshold [22]. Thus, drivers’ visual recognition distance for road obstacles must surpass the stopping sight distance. The stopping sight distance values required for safe driving under various vehicle speeds have been specified in standard guidelines. Furthermore, drivers’ visual recognition distance for road obstacles in different road lighting environments correlates with the visibility level (VL). The visibility level (VL) of a road surface obstacle indicates how much its actual luminance contrast exceeds the threshold contrast against the background, calculated with Equations (1) and (2). With a higher VL, the drivers’ visual recognition distance to the obstacles is also increased [23].

$$VL={\displaystyle \frac{C_A}{C_{th}}}$$

(1)

where $VL$ is the visibility level; $C_A$ is the actual luminance contrast between the obstacle and the road background; and $C_{th}$ is the threshold luminance contrast at which the obstacle is nearly visible against the background.

$$C_A={\displaystyle \frac{\left(L_o-L_b\right)}{L_b}}$$

(2)

where $L_o$ is the obstacle luminance, cd/m² and $L_b$ is the background luminance, cd/m².

Under bright–dark lighting conditions, when the driver recognizes obstacles in a dark environment from a bright environment, the high-luminance light entering the eye scatters to form an equivalent veiling luminance ($L_{veil}$) that uniformly overlays the retina. At this time, the eyes’ adapted luminance ($L_a$) is the sum of the obstacles’ background luminance ($L_b$) and the equivalent veiling luminance ($L_{veil}$), as shown in Equation (3). As the adaptive luminance in the drivers’ visual field increases, the contrast threshold required to detect a road target decreases, thereby improving the obstacles’ visibility level and extending the drivers’ visual recognition distance [24]. On the other hand, the equivalent veiling luminance in the bright–dark lighting environments attenuates the perceived contrast ($C_B$) of obstacles [25], thereby lowering the obstacles’ visibility level and shortening visual recognition distance, as shown in Equation (4) and Figure 2. Therefore, the magnitude of equivalent veiling luminance in bright–dark lighting environments directly influences the distance at which drivers can detect road surface obstacles.

$$L_a=L_b+L_{veil}$$

(3)

where $L_a$ is the adaptive luminance, cd/m² and $L_{veil}$ is the equivalent veiling luminance, cd/m².

$$C_B={\displaystyle \frac{\left(L_0+L_{veil}\right)-\left(L_b+L_{veil}\right)}{\left(L_b+L_{veil}\right)}}=C_A{\displaystyle \frac{L_b}{L_b+L_{veil}}}$$

(4)

where $C_B$ is the contrast under bright–dark lighting environments.

As the driver moves from the bright environment to the dark environment with a fixed visual field angle, the proportion of bright area in that field shrinks while the dark area grows, thereby altering the magnitude of the perceived equivalent veiling luminance. Adrian’s proposed method for calculating equivalent veiling luminance based on polar diagrams determines the luminance value by measuring the brightness of each sector in the polar diagram shown in Figure 3 and integrating these values, as detailed in Equation (5) [26,27]. Hu et al. simplified this model, demonstrating that the equivalent veiling luminance can be calculated as a weighted average of the luminance of each zone multiplied by its corresponding area fraction [28]. The luminance of each zone is determined by the design values of the bright–dark environments, while the area proportions depend on the spatial position of the driver or the small target within the bright–dark lighting environment.

$$L_{seq}=K{\displaystyle {\int }_0^{2\pi }L\left(\theta ,\varphi \right)}\mathrm{d}\varphi {\displaystyle {\int }_{{\theta }_1}^{{\theta }_2}\frac{\sin \theta \cos \theta }{{\theta }^2}}\mathrm{d}\theta$$

(5)

where $L_{seq}$ is the equivalent veiling luminance, cd/m²; $L\left(\theta ,\varphi \right)$ is the luminance value at the polar coordinate $\left(\theta ,\varphi \right)$, cd/m²; $\theta$ is the angle between the drivers’ line of sight and the glare source, ° or rad; $\varphi$ is the angle between the calculation point and the polar coordinate axis, ° or rad; and K is a constant: when $\theta$ is expressed in degrees, K = 10; when $\theta$ is expressed in radians, K = 0.003.

Through the above analysis, three key factors affecting drivers’ visual recognition distance for road obstacles in bright–dark lighting environments are identified: (1) the luminance of the dark environment, (2) the luminance of the bright environment, or luminance ratio of the bright–dark environments, and (3) the position of the obstacle (or the driver) within the bright–dark lighting environments.

3. Methods

To investigate how the bright–dark lighting environment inside the tunnel influences drivers’ detection distance, an experimental scheme was developed with three key factors. A combined objective–subjective methodology was adopted to conduct naturalistic driving small target recognition experimentally investigated variation patterns and characteristics of drivers’ recognition distance.

3.1. Experimental Scheme

The designed experimental scheme consists of the three key factors: the luminance of the dark environment, the bright–dark luminance ratio, and the position of the small target.

3.1.1. The Luminance of the Dark Environment and the Bright–Dark Luminance Ratio

In accordance with the Chinese Guidelines for Design of Lighting of Highway Tunnels (JTG/T D70-2-01-2014) [9], three interior zone luminances corresponding to the operating speed of 80 km/h were selected as the study conditions for the luminance of the dark environment, i.e., 1.5 cd/m², 2.5 cd/m², and 3.5 cd/m².

To cover a broader range of bright–dark luminance ratios, the Chinese guidelines require that the luminance of the diverging zone should be at least three times of the interior zone value. Therefore, it was extended to have ratios of 2, 3, 4, and 5 as the test conditions.

Combining the dark environment luminance and the bright–dark luminance ratios, 12 test conditions of the bright–dark lighting environment were obtained, listed in

Table 1. Test conditions of bright–dark lighting environment.

Number	Bright Diverging Luminance (cd/m2)	Dark Interior Luminance (cd/m2)	Bright–Dark Luminance Ratio
1	3.0	1.5	2
2	4.5	1.5	3
3	6.0	1.5	4
4	7.5	1.5	5
5	5.0	2.5	2
6	7.5	2.5	3
7	10.0	2.5	4
8	12.5	2.5	5
9	7.0	3.5	2
10	10.5	3.5	3
11	14.0	3.5	4
12	17.5	3.5	5

.

3.1.2. The Position of the Small Target

The midpoint between the last luminaire in the diverging zone and the first luminaire in the interior zone is taken as the bright–dark lighting environment boundary, i.e., the origin. With the origin (0 m) as the center, a 100 m test area is established, extending 50 m into the bright environment (negative) and 50 m into the dark environment (positive). The small target is placed every 10 m within the above mentioned test area, resulting in 11 positions. All positions of the small target are shown in Figure 4.

3.2. Experimental Road and Lighting Facilities

The experimental road is a straight road with a total length of 1000 m, with a maximum grade of 2.7%; the driver’s visibility of small targets on the pavement is unaffected by alignment conditions. The pavement material of the road is asphalt concrete.

Luminaires are symmetrically arranged on both sides with a spacing of 5 m; 100 W LED lamps are used in the diverging zone and 35 W LED lamps in the interior zone; all sources have a correlated color temperature of 4500 K and a color rendering-index of Ra = 70. Each lighting circuit is 500 m long, and brightness is regulated by a local centralized controller, which allows the road surface luminance to vary from 0 to 25 cd/m², as shown in Figure 5.

3.3. Small Target

Following the recommendations in CIE 2004, Guide for the Lighting of Road Tunnels and Underpasses [29], a 20 cm× 20 cm× 20 cm gray cube with a 20% surface reflectivity was used as the small target, as shown in Figure 6.

3.4. Instruments and Equipment

A non-contact multifunctional speedometer was used to collect the speed and distance of the vehicles in the tunnel, shown in Figure 7a. Meanwhile, it was also used to calibrate the characteristic positions inside the tunnel. The machine has a sampling frequency is 60 Hz; it has a speed range from 0 to 250.00 km/h with a resolution of 0.01 km/h; it also has a distance range from 0 to 9999.99 m with a resolution of 1 mm.

An imaging luminance meter, shown in Figure 7b, is used to measure road surface luminance inside the tunnel. It has its specifications: luminance range 0.05~4 × 10⁵ cd/m², accuracy better than ±3%, repeatability ≤ 1.5%.

3.5. Participants

To eliminate the influences of individual differences in age, gender, and cognitive level on the test results and ensure data reliability, recruited participant drivers must meet the specified requirements. According to the data released by the Ministry of Public Security of the People’s Republic of China, as of the end of 2024, drivers aged between 26 and 50 accounted for more than 70% of the total number of drivers. Therefore, to ensure that the participant drivers represent the vast majority of drivers, all participant drivers recruited in the preparation phase of the experiments were aged between 26 and 50. In accordance with the guidelines for human factors experimental design and sample size selection [30,31], the sample size of the participant drivers was set at 24. The basic information of the participants is listed in

Table 2. Details of the participant drivers in the tests.

Number of Participants	License Type	Age (26–45 years)		Driving Experience (≥2 years)
		Mean	SD	Mean	SD
Male (n = 16)	C1	34.56	7.65	11.75	6.12
Female (n = 8)	C1	32.17	7.15	8.67	5.53

Notes: “n” represents the sample size, “SD” means the standard deviation, and “C1” refers to a type of driver’s license in China that authorizes the holder to drive passenger vehicles with up to 9 seats and a maximum total weight of 4500 kg.

. Each participant holds a valid C1 license with at least two years of driving experience. Meanwhile, all of them had uncorrected or corrected visual acuity of ≥4.9, and had abstained from alcohol and medication for the past week (were well-rested and exhibited normal reaction times).

3.6. Driver Subjective Evaluation Rating Scale

Based on the visual task of whether drivers can timely identify road surface targets during driving, the detection level for small targets is classified into three grades: clear, general clear, and blurred; the corresponding descriptions are given in

Table 3. Rating scale for small target detection descriptions.

Level	Visibility	Statement	Qualified
1	Clear	Clear outline and accurate identification	Yes
2	General clear	General clear outline	Yes
3	Blurred	Blurred and unclear outline	No

. When the detection level of small targets in the driver’s visual field reaches Level 2, it is regarded as the critical state satisfying safe visual recognition [22].

3.7. Experimental Procedure

To collect the visual recognition distance of small targets that meet drivers’ subjective recognition needs under different experimental conditions, the specific experimental procedure is defined as follows. Before the experiment begins, participants are trained to familiarize themselves with and clearly understand the subjective recognition scale shown in Table 3. The test personnel adjust the luminance values of the dark environment and the bright environment according to the test conditions specified in Table 1 and place the small targets at the positions shown in Figure 4. When the participant receives the start signal at the upstream of the test road, they begin driving, accelerate to 80 km/h, and maintain this speed for continued travel. The participant detects the small target on the road, and once the target’s visibility reaches Level 2 in the rating scale in Table 3, they prompt the test personnel in the vehicle immediately while maintaining the vehicle’s speed and continuing travel. The test personnel in the vehicle will immediately mark T₁ using a non-contact speedometer upon receiving the driver’s notification. The test vehicle continues moving forward. When it passes the small target, the staff test personnel in the vehicle will mark T₂ using a non-contact speedometer, recording the distance between the two markers. The participant is then replaced or the bright–dark lighting environment is adjusted and the above steps are repeated until all trials are completed. The experimental procedure flow is shown in Figure 8.

4. Results

A total of 3168 data sets of drivers’ visual recognition distance for the small target were collected. The 218 test cases of drivers’ visual recognition distance for small targets were significantly different from that of other drivers under identical lighting conditions due to the interference of construction vehicle headlights in the tunnel. After analyzing and removing outliers, 2950 valid data sets remained.

4.1. Variation Patterns of Visual Recognition Distance and Influencing Factors

The visual recognition distances data of 24 participants under 132 experimental conditions were analyzed separately for normality using the Shapiro–Wilk (S-W) test. The results of p-value were between 0.13 and 0.20, all greater than 0.05, which indicates that the data followed a normal distribution. The standard deviations (SDs) of data sets ranged from 3.29 to 5.42 m, which satisfies the condition for using the mean as the representative value.

Based on the mean visual recognition distances of drivers for small targets under different experimental conditions obtained from data processing, the relationship between the position of the small target and the visual recognition distance is plotted in Figure 9. To determine whether the position of small target change affects drivers’ visual recognition distance, a one-way ANOVA was conducted on visual recognition distance data within the range of −50 to 0 m. The result is a p-value below 0.05 and indicates that, under the same conditions of the bright–dark luminance ratio and dark environment luminance but with different positions of the small target, the mean visual recognition distance of the participants shows significant differences as the position of the small target changes. Similarly, within the range of 0~50 m, the mean visual recognition distances of the drivers also show significant differences. Therefore, the placing position of the small target affects the drivers’ visual recognition distance.

According to Figure 9, as the position of the small target changes from −50m to 0m, the mean visual recognition distance of drivers for the small target under the identical conditions gradually decreases. There is a negative correlation between the position of the small target and the visual recognition distance (r = −0.942). As the position of the small target changes from 0 m to 50 m, the mean visual recognition distance of drivers for the small target under identical conditions gradually increases. There is a positive correlation between the position of the small target and the visual recognition distance (r = 0.852). From the analysis of the variation pattern and correlation between the placing position of the small target and the visual recognition distance, it can be concluded that the closer the small target is placed to the 0m position (the origin) at the boundary between the bright and dark lighting environments, the shorter the visual recognition distance of the small target by the drivers.

Based on the mean visual recognition distance data of drivers for the small target under identical other conditions, the variation relationship between the drivers’ visual recognition distance and the bright–dark luminance ratio was plotted, as shown in Figure 10. A one-way ANOVA was conducted on the mean visual recognition distance of the small target under the same dark environment luminance but different luminance ratios between the bright and dark environments. The result is a p-value below 0.05 and indicates that the luminance ratio has a significant impact on the drivers’ visual recognition distance. As can be seen from Figure 10, as the bright–dark luminance ratio increases, the visual recognition distance of drivers for the small target decreases. The results of the Pearson correlation test show a negative correlation between the bright–dark luminance ratio and the visual recognition distance (r = −0.569).

Based on the mean visual recognition distance data of drivers for the small target under identical other conditions, the variation relationship between the drivers’ visual recognition distance and the dark environment luminance was plotted, as shown in Figure 11. Under the condition of the same placing position of the small target and the same bright–dark environment’s luminance ratio but a different dark environment luminance, a one-way ANOVA was conducted on the mean visual recognition distance of the small target by the participant drivers. The result is a p-value below 0.05 and indicates that the dark environment luminance has a significant impact on the drivers’ visual recognition distance. As can be seen from Figure 11, as the luminance of the dark environment increases, the visual recognition distance of drivers tends to increase. The results of the Pearson correlation test indicate a positive correlation between the luminance of the dark interior and visual recognition distance (r = 0.464). From the analysis of the visual recognition distance data, it can be concluded that, under the condition of a constant placing position of the small target and a constant luminance ratio of bright–dark environmental conditions, as the luminance of the dark environment increases, that is, as the overall luminance of the bright–dark lighting environments increases, the visual recognition distance of drivers for the small target increases, and their visual recognition ability correspondingly improves.

4.2. Visual Recognition Distance Fitting Model

To ensure the safe driving recognition needs of drivers, the visual recognition distance of drivers for small targets on the road surface needs to be greater than the safe stopping sight distance. From the above analysis, it is shown that the bright–dark luminance ratio, the luminance of the dark environment, and the placing position of the small target all affect the driver’s recognition distance. When the luminance ratio and the luminance of the dark interior are constant, the visual recognition distance of the small target by the driver is the shortest when the small target is placed at the boundary (the origin) between the bright and dark lighting environments. When the minimum recognition distance in the bright–dark lighting environments is greater than the stopping sight distance, it indicates that all bright–dark lighting conditions meet the drivers’ visual recognition distance and safe driving requirements.

In order to determine whether different bright–dark lighting environments meet the drivers’ recognition distance requirements for safe driving and whether they affect driving safety, based on the experimental data, regression analysis was used in Origin 2024 software to establish a bivariate quadratic fitting model between the visual recognition distance of the small target placed at the boundary (the origin) between the bright and dark lighting environments and the independent variables of the bright–dark luminance ratio and the luminance of the dark interior shown in Equation (6). The correlation test results of the fitting model show that the determination coefficient is R² = 0.98, which indicates a good fit of the model.

$$D=302.1-123.68L_{\mathrm{in}}-36.94C+24.41L_{\mathrm{in}}^2+1.36C^2+4.83L_{\mathrm{in}}C$$

(6)

where D is the visual recognition distance; L_in is the luminance of the dark interior zone (1.5 ≤ L_in ≤ 3.5 cd/m²); and C is the luminance ratio between the bright and dark lighting environments (2 ≤ C ≤ 5).

Based on the visual recognition distance fitting model, the variation trend of the visual recognition distance of drivers for small targets at the boundary between bright and dark lighting environments was plotted as the luminance ratio of bright–dark areas (changed from two to five) and the luminance of the dark interior (changed from 1.5 cd/m² to 3.5 cd/m²), as shown in Figure 12. As it can be seen from Figure 12, when the luminance of the dark environment is different, the luminance ratio between the bright and dark environments needs to be different in order to ensure that the visual recognition distance of drivers in the bright–dark lighting environments reaches the safe stopping sight distance. The stopping sight distance required for the operating speed of 80 km/h used in the test is 110 m. When the luminance of the dark environment is selected as 1.5 cd/m², 2.5 cd/m², or 3.5 cd/m²_, according to the specifications, the threshold values of the bright–dark luminance ratio that meet the drivers’ safe visual recognition distance requirements are calculated to be 2.29, 1.59, and 3.97, respectively.

5. Discussion

The variation patterns and characteristics of drivers’ visual recognition distance under the influence of factors, i.e., the bright–dark environments luminance ratio, the luminance of the dark environment, and the position of small targets in the bright–dark lighting environments of the diverging zone in an interchange within a tunnel were investigated in this study. A fitting model was established between the visual recognition distance and the bright–dark luminance ratio, the luminance of the dark environment. The threshold value of the luminance ratio between the bright environment (diverging zone) and the dark environment (interior zone) was analyzed to meet the requirement of safe driving stopping sight distance. The research results show that when the vehicle operating speed inside the tunnel is 80 km/h, the threshold of luminance ratio is less than that specified by the regulations, i.e., three times when the luminance of the interior zone of the tunnel is set as 1.5 cd/m² or 2.5 cd/m². If the diverging zone lighting is set according to the specification (luminance ratio as three times), drivers’ visual recognition distance for road obstacles at the boundary between the diverging zone and interior zone will be shorter than the safe stopping sight distance, which creates a traffic safety hazard.

By comparing drivers’ visual recognition distance in bright–dark lighting environments with the safe stopping sight distance, an upper-limit threshold for the bright–dark luminance ratio to meet the visual demands for safe driving was proposed in this study. This differs from the specification’s lower-limit requirement that the diverge area luminance should not be less than three times the interior zone luminance, which is intended to increase and safeguard driving safety by increasing the diverge zone luminance. Experimental results show that when the luminance ratio in the bright–dark lighting environments increases, drivers’ recognition distance decreases. When the luminance of the interior zone remains constant, a higher ratio enlarges the luminance difference between the bright and dark environment, which also increases the equivalent veiling luminance within the driver’s visual field. This reduces the visibility of small targets and thus shortens visual recognition distance [24]. Therefore, establishing an upper-limit threshold for the bright–dark luminance ratio is recommended to ensure adequate stopping sight distance for drivers at interchange diverging zone within tunnels and to enhance traffic safety.

On the other hand, the luminance of the dark environment in the bright–dark lighting environments also affects drivers’ safe visual recognition distance. Existing studies indicate that, when the luminance ratio of the bright–dark lighting environments remains constant, synchronously increasing the luminance of both the diverging zone and the interior zone raises the adaptive luminance of the driver’s visual field, and proportionally enlarges the luminance difference between the small target and its background, thereby improving the driver’s visual recognition distance and ability for objects ahead [12]. Likewise, to achieve the safe stopping sight distance, the threshold of the bright–dark luminance ratio that ensures safe driving varies with the luminance of the dark environment.

In the tunnel threshold zone characterized by a bright–dark lighting environments, Hu et al. [15] examined lighting quality and conducted naturalistic driving visual recognition distance tests for small targets. They found that visual recognition distance is positively correlated with the threshold zone luminance reduction factor (the reciprocal of the luminance ratio). To satisfy the safe stopping sight distance for an 80 km/h operating speed, a threshold zone luminance reduction factor of 0.045 (corresponding to a luminance ratio of about 22) was derived. Compared with the bright–dark lighting environments in the interchange diverging zone of a highway tunnel, the overall luminance level in the tunnel threshold zone is higher; the findings indicate that a larger luminance ratio can be set to ensure driving safety. This indirectly confirms that, as the luminance of the dark environment in the bright–dark lighting environments increases, the allowable threshold value of the luminance ratio for safe driving also increases. Comparative analysis of the two bright–dark lighting environments reveals that, although the patterns governing the effects of visual recognition distance factors are similar, the numerical values differ significantly. The findings of this study supplement and refine the threshold of the bright–dark environments luminance ratio for interchange diverging zones in highway tunnels, where the overall luminance is low, yet a luminance difference exists between the bright (diverging zone) and dark (interior zone) environments. These results provide valuable guidance for establishing lighting conditions that ensure safe stopping sight distances and enhance driving safety in such highway tunnels.

Due to field-test constraints, this study examined naturalistic driving visual recognition distance only at 80 km/h. Previous findings show that visual recognition distances for road obstacles in tunnel threshold zone decrease with increasing speed [14]. Therefore, the influence of vehicle speed on visual recognition distance in interchange diverging zone within tunnels requires further investigation. Moreover, the color temperature of the light source also affects drivers’ visual recognition distance for road obstacles [32]. Since the on-site luminaires were all a fixed color temperature (4500 K), this study only examined recognition patterns under that specific color temperature. How recognition distance varies under other color temperatures source remains to be studied in future research.

6. Conclusions

Based on an analysis of the key factors affecting driver visual recognition ability in bright–dark lighting environments, the naturalistic driving tests using small targets to investigate the variation patterns and characteristics of driver visual recognition distance under the bright–dark lighting conditions between the interchange diverging zone and the interior zone of tunnels were conducted in this study.

In a bright–dark lighting environments, factors such as the position of a small target, the bright–dark environments luminance ratio, and the luminance of the dark environment all significantly affect drivers’ visual recognition distance. When the bright–dark lighting environments remain unchanged and the small target is placed at the boundary, its impact on driver visual recognition distance is the greatest. If the luminance ratio is kept constant while the overall luminance of both environments increases, the visual recognition distance improves. Conversely, if the dark environment luminance stays the same and the bright–dark environments luminance ratio rises, the visual recognition distance decreases.

The developed bivariate regression model relating to the dark environment luminance, bright–dark environment luminance ratio, and visual recognition distance of small targets at the bright–dark boundary shows that the luminance ratio threshold required for safe stopping sight distance varies with changes in interior zone luminance.