Road zoning in the Fergana Valley based on the level of traffic safety

2.1. Data sources

The analysis uses official statistical records on vehicle ownership, traffic accidents, and road network characteristics from regional authorities and national datasets [1-5]. Additional data on traffic flow, pavement condition, and climatic parameters were collected on major highways, including A373, 4R112, 4R115, 4R126, and 4R140.

2.2. Traffic flow measurements

Traffic flow characteristics – speed, intensity, and density – were measured using the “Trassa” mobile road laboratory. The system recorded vehicle classification, hourly flow intensity, and mean operating speed. Measurements were conducted on both two- and four-lane roads under typical daytime traffic conditions. Flow density $q$ was calculated as:

where $N$ is traffic intensity (veh/h) and $V$ is traffic speed (km/h).

2.3. Pavement and roadway conditions

Pavement condition, geometric features, lighting, pedestrian infrastructure, and roadside safety elements were assessed according to national standards. The factors most strongly associated with accident probability were identified by comparing road condition data with historical accident patterns.

2.4. Accident data and classification

Traffic accidents from the last ten years were categorized by type, cause, time of occurrence, and road category. The distribution of accident types and their underlying causes was used to determine the most hazardous sections and contributing factors.

2.5. Statistical analysis

All experimental measurements were repeated three times to ensure consistency. Standard deviation (SD) and expanded uncertainty ($U=k·SD$, $k=$ 2 for 95 % coverage) were computed. Regression analysis was used to evaluate the relationship between traffic speed, flow density, and safety coefficient. Confidence intervals (95 %) were applied to validate trend stability.

In total, 48 measurement points were collected for two-lane roads and 52 points for four-lane roads. Outliers were removed using a 1.5·IQR criterion. The average standard deviation of repeated measurements was 1.7 % for speed, 2.3 % for density, and 2.0 % for the safety coefficient. The regression analysis was performed using a linear model, and the reliability of the fitted trends was validated through R², residual distribution, and confidence intervals.

2.6. Composite safety indicator (CSI)

To evaluate road safety levels, a Composite Safety Indicator was developed by integrating the relative accident rate, traffic-flow characteristics, road condition, and climatic factors. The indicator enables ranking of districts from highest to lowest safety performance and forms the basis for zoning.

3.1. Vehicle distribution and road network characteristics

The Fergana Valley contains 995,821 vehicles, of which passenger cars represent 86.4 % and trucks 11.4 % (Table 1). Variations in vehicle concentration per kilometer of road indicate uneven transport loading, with the regional average reaching 21.05 vehicles/km. Figs. 1 and 2 illustrate the geographical layout and vehicle distribution, showing that districts with higher density experience greater exposure to traffic risk, consistent with international findings on crash probability and network load [7], [11], [15].

Fig. 1Road network map of the Fergana Valley

3.2. Accident trends and risk patterns

In 2022-2023, Uzbekistan recorded about 10,000 traffic accidents annually, with 31-35 % occurring in the Fergana Valley (Fig. 3). This confirms the region’s classification as a high-risk (“red zone”) area.

Analysis over the last decade shows that pedestrian collisions (44.9 %), vehicle collisions (31.2 %), and bicycle collisions (11.5 %) dominate crash types (Fig. 4). The predominance of pedestrian crashes reflects insufficient lighting, missing fences, and high mixed-traffic interaction, aligning with global studies [12].

Fig. 2Analysis of the distribution of vehicles and road network density by regions of the Fergana Valley

Fig. 3Distribution of traffic accidents across regions of the Republic in 2023 and their comparative analysis with 2022

Fig. 4Distribution of road traffic accidents per 1 km of road in the Fergana Valley by types

3.3. Causes of accidents

Key contributing factors include:

1) 14 % – absence of median barriers.

2) 10 % – insufficient pedestrian barriers.

3) 9 % – inadequate lighting.

4) 14 % – speeding.

5) 17 % – failure to yield to pedestrians.

6) 8 % – dangerous overtaking.

Accidents occur most frequently on national and urban roads, while A373 and 4R112 show the highest accident density, making them the primary study corridors.

3.4. Traffic flow measurements

Traffic-flow analysis using the “Trassa” mobile laboratory revealed an inverse relationship between speed and density on two- and four-lane roads. The measured speed–density curves (see Fig. 5) show that increasing density reduces maneuvering space and elevates psychological load on drivers, leading to higher accident likelihood. Heavy vehicles contribute disproportionately to flow saturation, which agrees with earlier research on pavement and safety degradation under truck loads [6], [15].

3.5. Speed-density-safety relationship

Fig. 5 shows the integrated dependence between traffic speed, traffic flow density, and the safety coefficient (Ksc) for two-lane and four-lane roads.

The regression trends confirm that:

1) A 10 km/h decrease in traffic speed leads to an increase in density by +8.6 veh/km (2-lane) and +13.5 veh/km (4-lane).

2) A 10 km/h increase in traffic speed causes the safety coefficient to rise by +0.09 (2-lane) and +0.08 (4-lane).

These results indicate that speed stabilization helps prevent density escalation and significantly improves the operational safety level, especially under mixed-traffic conditions.

A detailed explanation of the unusually linear distribution of data points is provided below to address measurement reliability and model validity.

Fig. 5Integrated relationship between traffic speed, traffic flow density (left Y-axis) and safety coefficient Ksc (right Y-axis) on two-lane and four-lane roads. Linear regression lines are shown together with 95 % confidence intervals

3.6. Explanation and validation of linear trends

The observed linearity of the speed-density and speed-safety coefficient relationships in the range of 40-100 km/h is primarily due to the controlled measurement conditions. All measurements were performed on straight, uniform road segments, during daytime, under dry pavement and stable weather, and with nearly constant traffic composition. Under such stabilized traffic conditions, fluctuations in headway, braking behavior, and lateral maneuvering are significantly reduced, which results in quasi-linear trends that resemble the sub-critical region of the classical fundamental diagram.

To verify the stability of the measurements, each point was recorded three times and the standard deviation values remained within 1.7-2.5 %. Regression diagnostics showed high coefficients of determination ($R^2=$ 0.86-0.91), and residual analysis did not reveal clustering or heteroscedasticity. These results confirm that the fitted linear models are statistically sound for the selected measurement interval.

However, it must be emphasized that such linearity may not fully characterize congested traffic, peak-hour conditions, or non-uniform flow influenced by intersections, weather changes, or high truck intensity. Therefore, the presented linear relationships apply specifically to controlled measurement conditions.

3.7. Composite safety indicator (CSI) and zoning

A Composite Safety Indicator was developed by integrating accident rate, traffic-flow parameters, road condition, and climatic influences. Districts were ranked “from largest to smallest” CSI value to identify zoning boundaries. The results indicate four safety zones:

1) Zone I (0.00–0.77): Kokand.

2) Zone II (0.77–0.81): Fergana.

3) Zone III (0.81–1.16): Andijan.

4) Zone IV (1.16–1.22): Namangan, Pop.

Fig. 6 presents the zoning map.

Fig. 6Road zoning map based on the level of road traffic safety

This classification allows decision-makers to prioritize investments in safety infrastructure and identify high-risk corridors requiring immediate intervention.

3.8. Practical implications and limitations

The study shows that accident risk in the region is driven by a combination of traffic load, insufficient infrastructure, and climate-induced factors. Improving lighting, installing barriers, regulating heavy-vehicle movement, and enhancing pavement quality are critical measures for raising road-safety levels.

The proposed CSI-based zoning methodology may be adapted for other regions with similar transport environments.

This classification clearly identifies areas requiring urgent safety interventions and supports the prioritization of infrastructure improvements.

The linear speed-density-safety relationships obtained in this study reflect controlled daytime traffic and may not directly apply to saturated, peak-hour, nighttime, or weather-affected conditions. Future research will expand the dataset using real-time monitoring and include non-linear traffic states to ensure broader applicability of the model.