Factors Contributing to Median Encroachments and Cross-Median Crashes (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

37 S E C T I O N 4 This section of the report describes the field data collection methodology and presents the results of the interdisciplin- ary field reviews conducted as part of the current research. The section includes an overview of the interdisciplinary field review approach, a description of the site selection activities, a description of the interdisciplinary investigations, and a summary of the results obtained. 4.1 Overview of Interdisciplinary Field Review Approach Interdisciplinary field reviews were conducted for more than 40 divided highway sites with high frequencies of median-related crashes to identify the most common con- tributing factors to those crashes. The sites with high fre- quencies of median-related crashes were identified using the network screening algorithms developed to identify sites with high crash frequencies in the Safety Analyst software tools for safety management of specific highway sites (51). The selected sites all had relatively high frequencies of median- related crashes, but were not necessarily, in all cases, the sites with the highest median-related crash frequencies. Copies of police crash reports, including the officer’s narrative descrip- tion of the crash, were obtained for all median-related crashes that occurred at each site during a period of 3 to 5 years. An interdisciplinary team consisting of a highway traffic engi- neer specializing in crash analysis and highway geometric design, and a human factors engineer specializing in driver behavior studies, reviewed all of the crash reports and visited each site in the field to review the geometric design and traf- fic control features of the site, the characteristics of the spe- cific crash locations, and observable traffic patterns. Based on these reviews, the interdisciplinary team classified the work- load level for drivers traversing the site, and identified specific roadway factors that explain the driver workload level and contributed to the occurrence of the crashes. In subsequent tasks, described in Section 5 of this report, the research team investigated whether the contributing factors observed at the selected field sites are overrepresented in median-related crashes at these sites (i.e., whether the identified factors are, in fact, associated with increased likelihood of median-related crashes). Section 6 of the report identifies countermeasures that address the contributing factors and, therefore, might be used to reduce median-related crashes. 4.2 Site Selection for Interdisciplinary Field Reviews Sites for the interdisciplinary field studies were sought in four states: California, Missouri, Ohio, and Washington. Network screening analyses were conducted for divided high- ways on the entire state highway systems of these states to identify sites with high median-related crash frequencies. Separate analyses were conducted for rural freeways, other rural divided highways (nonfreeways), and urban freeways. The network screening analyses for divided highways in each state were conducted with network screening procedures based on both the peak-searching and sliding-window algo- rithms developed by Harwood et al. (51) in FHWA research for use in the Safety Analyst software tools for safety manage- ment of specific highway sites. The network screening analy- ses were not conducted with the Safety Analyst software, but rather by using the published network screening algorithms programmed in the SAS software package. Roadway, traffic volume, and crash data for these analyses were obtained for California, Ohio, and Washington from the FHWA HSIS, and for Missouri from the MoDOT Transportation Management System (TMS). The network screening analyses identified sites with high fre- quencies of median-related crashes that were sorted in descend- ing order of median-related crash frequency per mile per year. Each site consisted of a directional divided highway segment Interdisciplinary Field Reviews

38 (i.e., the roadway in one direction of travel on a divided high- way), generally at least 1.6 km (1.0 mi) in length. The inter- disciplinary field study sites were selected from among these high crash frequency sites. The research team sought a min- imum of 10 study sites in each state. A total of 47 sites were selected for evaluation, as follows: • California 10 sites, • Missouri 12 sites, • Ohio 13 sites, and • Washington 12 sites. These sites do not necessarily represent the sites with the very highest median-related crash frequencies in each state. This is due to the fact that geographic locations of the sites were considered in site selection so that it would be feasible for the interdisciplinary team to visit all sites in a given state during a 1-week period. Sites were selected so that each site experienced at least 10 median-related crashes over a 5-year period. The majority of the sites were located on rural free- ways. Finally, selections were made to ensure that the overall set of sites included some rural divided highways (nonfree- ways), some urban freeways, and a range of median types, median widths, and terrain types. 4.3 Interdisciplinary Field Review Procedure The object of the interdisciplinary field reviews was to identify factors associated with median encroachments, espe- cially those associated with the initiation of cross-median crashes at sites with higher-than-expected crash frequen- cies identified through a screening of all potential sites. This review addressed both engineering and human factors con- siderations and was conducted by an interdisciplinary team that included engineers from MRIGlobal and human factors specialists from Human Factors North. The interdisciplinary review considered geometric design, traffic control, and traffic operational, weather, and driver condition factors and involved a review of hard-copy police crash reports, video logs of crash sites, and aerial photo- graphs prior to field visits to crash sites. These office reviews were performed first and led to a decision as to which sites were most suitable for field visits. The interdisciplinary team visited all sites in a particular state on a single trip and spent a sufficient amount of time at each site to collect geometric data, observe traffic behav- ior at various locations, videotape a drive-through of the site from the driver’s perspective, and—when appropriate— collect speed data. The specific nature of the reviews con- ducted at each site was based on the crash history of the site; however, the general plan for the field investigations follows. The interdisciplinary field investigations included consider- ation of four elements. • Driver tasks and information requirements, • Traffic operations, • Roadway and roadside design, and • Environmental considerations. A critical aspect of the field investigation was the system- atic examination of the driver’s task in negotiating the loca- tion under investigation. Driver tasks involve all the potential tasks that a driver might make through a site. On a freeway, this could include sign reading, merging, lane changing, passing, slowing, and exiting. Because a majority of colli- sions involved human error, the team took opportunity to drive through the site multiple times to understand the level of workload required of a driver. The investigators were ade- quately prepared with an understanding of collision history and patterns for the site, as well as potential performance issues. It was during these runs that the human factors exper- tise was the focal point of the investigation. The goal of the drive-through was to • Experience the site from the perspective of the highway user, keeping in mind unfamiliar users. • Identify highway elements and roadside features that may be contributing to driver error—in particular, to note usual highway designs that may violate driver expecta- tions (e.g., left exits on freeways, traffic queuing at an exit ramp, drivers entering the freeway at low speeds because of short acceleration lanes, drivers changing lanes at the last second because of inadequate bullnose-to-bullnose separation or inadequate placement of guide signs, etc.); features, such as an usually sharp curve, that are different from what the driver might expect given experience of the road upstream; and cues that may mislead the driver about the actual road alignment (e.g., tangential exit ramp within mainline curves). • Check adequacy of sight distances to hazards such as lane drops, lane splits, bridges, tunnels, and work zones. • Check sufficient warning is given for hazards, especially at night, and especially for hazards featured in past collisions. • Check that signs are conspicuous, especially in cluttered backgrounds; easily understood; legible at the point at which the driver needs the information to make lane changes or other maneuvers; and spaced to allow easy reading. • Check that markings are clearly visible and likely to be understood by unfamiliar drivers. • Determine whether driver attention is distracted at critical points when it should be focused elsewhere (e.g., a bill- board to the right, when the driver is approaching an inter- change with the potential of slowing traffic ahead).

39 These interdisciplinary site reviews at locations with a high frequency of median-related or cross-median crashes lead the researchers to identify factors related to high driver- workload conditions, such as the presence of interchange ramps, horizontal curves, lane drops, or combinations of these factors. These reviews were conducted in an unbiased manner, and the study team was careful that their findings were not shaped by preconceptions as to which factors might prove to be important. This approach gave an opportunity for unanticipated factors to come to light. Based on the contributing factors identified (or the absence of factors), the field visits, and review of the videos, the inter- disciplinary review team classified the driver workload at each site into one of seven categories based on task difficulty in relation to road design, as follows: • Very low driver workload—straight and level, no entrances or exits, no lane drops; • Low driver workload—generally straight or gentle curves, gentle grade, may include exit to/entrance from rest area; • Low-to-moderate driver workload—generally straight or gentle curves, some grade, may include bridges or an exit to/entrance from a rest area, widely separated interchanges; • Moderate driver workload—gentle curves, moderate grade, compound curves, interchanges more than 2 miles apart, lane additions or lane drops, bridges; • Moderate-to-high driver workload—gentle curves, mod- erate grade, compound curves, interchanges 1.5 to 2 miles apart, entrances with short acceleration lanes, lane addi- tion or lane drops, bridges; • High driver workload—gentle curves, moderate grade, compound curves, interchanges less than 1.5 miles apart, lane additions or lane drops, difficult merging conditions (e.g., merging or curves or grades, or with short accelera- tion lanes), bridges; and • Very high driver workload—sharp curves, steep grades, interchanges less than 1 miles apart, compound curves, lane additions or lane drops, bridges, inappropriate super- elevation, entrance or exit ramps with sharp curves. In selecting the appropriate category, attention was paid not only to which geometric design elements were present, but their number, combination, and severity. Road condition also contributes to driver workload, with wet, snowy, and icy roads demanding more attention due to a lower coefficient of friction. Similarly, dense traffic will contribute to increased workload. With all these factors in mind, a global estimate was made of workload severity. Finally, it should also be noted that driver workload is a product of not only task difficulty, but also of driver effort and arousal. An easy task may be more difficult for an older, impaired, unfamiliar driver in comparison to a difficult task • Consider work zone issues (signing, marking, lane clo- sures, and lane re-alignments) that may have contributed to past collisions and may contribute to future ones. The team used a head-mounted audio and video recorder, as show in Figure 4-1, to record the site from the perspective of the driver. A video recording of the location was made to create a permanent record of each study site for review in the office. The video recording included a time and date stamp and a view of the roadway as seen from the driver perspec- tive (recorded with a camera attached to the frame of glasses worn by the driver). At the beginning of filming each section of roadway, the observer noted the following: • Site and location identifications, • Direction of the approach, and • Numbers of collisions associated with the approach. During filming of each section of roadway, the investigat- ing team also noted the following: • Sight distance to hazards and • Legibility distance of signs and markings. The verbal commentary that accompanied the video record- ing was important because not all signs, signals, and mark- ings are as visible on the video as they are to the human eye. Among other observations, the team noted when various hazards can be seen by the naked eye, and when signs are first legible. Figure 4-1. Head-mounted audio and video recorder.

40 Roadway Type • Rural freeways: 69.9 miles (72 percent); • Rural divided nonfreeways: 11.1 miles (11 percent); and • Urban freeways: 16.4 miles (17 percent). Terrain Type • Mountainous: 6.2 miles (6 percent) and • Level/rolling: 91.2 miles (94 percent). Median Type • Traversable: 11.6 miles (12 percent); • Nontraversable: 10.2 miles (10 percent); • Barrier: 75.6 miles (78 percent); – Guardrail: 4.3 miles (4 percent); – Concrete Barrier: 28.0 miles (29 percent); and – Cable Barrier: 43.3 miles (45 percent). 4.5 Interdisciplinary Field Review Results Table 4-2 identifies the classification of driver workload made for each site by the interdisciplinary review team and the contributing factors on which that classification was based. Tables 4-3 and 4-4 present individual contributing factors and combined contributing factors, respectively. No contributing factors to increasing workload were noted for roads that were essentially level and straight and without interchanges. Table 4-5 presents a frequency distribution of contributing factors that includes both individual and combined contrib- uting factors. Table 4-6 presents a frequency distribution for the individual contributing factors. 4.6 Interdisciplinary Crash Review Results Table 4-7 summarizes the median-related crash frequen- cies for the interdisciplinary field review sites, based on crash data for a 5-year period. Median-related crashes are defined in this study as crashes in which one or more of the involved vehicles left the roadway and entered the highway median. Crashes have been classified into categories based on crash initiation type (single-vehicle loss of control vs. vehicle inter- action) and pavement surface condition (dry conditions vs. wet conditions). 4.6.1 Crash Initiation Type Crash initiation type was categorized based on review of police crash reports, including review of the investigating officer’s narrative description of the crash. Single-vehicle loss-of-control crashes are those in which a vehicle left the roadway for reasons that appear unrelated to the action of any for a middle-aged, sober, and familiar driver. We assume that generally there are equal numbers of such drivers exposed to easy and difficult driving tasks, although this is not always true. Younger drivers do more of their driving at night as com- pared to middle-aged drivers, and older drivers do less of their driving in high-volume, poor traffic conditions as compared to middle-aged drivers. Generally, however, it is reasonable to assume that there are similar proportions of low-effort, low- arousal drivers on various sections of roadway. In addition to driving through the site multiple times, the team observed traffic operations from a convenient vantage point and took geometric field measurements for important elements, including grade, shoulder width, ramp spacing, and curve radii for locations and elements that could not be mea- sured in the office using aerial photographs or project plans. When a feature of potential interest—such as an interchange ramp or a horizontal curve—was found to be present at a site, the team took time to study that feature in detail to deter- mine whether and how it might be contributing to the initia- tion of median encroachments that may potentially lead to cross-median crashes. Depending on the nature of the loca- tion and its crash history, this review focused on geometric design elements, design dimensions, spacing between adja- cent design elements, traffic operational conditions (includ- ing time-of-day variations), signing, marking, speed zoning, speed transitions, lighting, pavement conditions, sun angle, and surrounding development. An important aspect of the interdisciplinary review was looking for evidence of median encroachments that did not result in reportable crashes. Such evidence may have included damaged roadside hardware, tire marks, or tire tracks. These “encroachment indicators” can furnish a qualitative idea of the encroachment level at a site without a quantitative mea- surement of encroachment frequency. 4.4 Site Characteristics for Interdisciplinary Field Review Sites Table 4-1 presents a summary of the characteristics of the 47 interdisciplinary field sites, including state, route, county, direction of travel, length, roadway type, median type, median width, and directional annual average daily traffic volume (AADT). The overall distribution of site characteristics is as follows: • Total length: 97.4 miles; • Average site length: 2.1 miles (range: 0.7 to 10.7 miles); • Directional AADT: Range: 7,500 to 229,000 vehicles per day; and • Median width: Range: 16 to 160 feet.

41 Site number County Route Direction of travel Length (mi) Roadway type Median typea Median width (ft)a Directional AADT (veh/day) California Sites CA01 Merced I-5 NB 2.0 Rural Freeway Traversable 75 39000 CA02 Merced I-5 NB 2.0 Rural Freeway Guardrail 80 39000 CA04 Sacramento I-5 NB 2.0 Rural Freeway Traversable 70 58000 CA05 Shasta I-5 NB 1.0 Rural Freeway Concrete Barrier 40 61000 CA07 Shasta I-5 NB 2.0 Urban Freeway Traversable 30 61000 CA10 San Joaquin SR 99 NB 1.0 Rural Freeway Concrete Barrier 20 75000 CA11 San Joaquin SR 99 NB 1.0 Urban Freeway Concrete Barrier 20 72000 CA13 San Joaquin SR 99 NB 1.0 Urban Freeway Concrete Barrier 10 67000 CA14 San Joaquin SR 99 NB 1.0 Rural Freeway Concrete Barrier 10 67000 CA15 San Joaquin SR 99 NB 1.0 Rural Freeway Concrete Barrier 10 64000 Missouri Sites MO01 Franklin I-44 EB 2.0 Rural Freeway Cable Barrier 30 15000 MO02 Franklin I-44 EB 3.0 Rural Freeway Cable Barrier 30 17000 MO03 Johnson US 50 EB 2.0 Rural Divided Nonfreeway Traversable 100 7500 MO04 Johnson US 50 EB 3.6 Rural Divided Nonfreeway Traversable 50 7000 MO05 Lafayette I-70 EB 10.7 Rural Freeway Cable Barrier 30 20000 MO06 Boone/Callaway I-70 EB 2.0 Rural Freeway Cable Barrier 30 17000 MO07 Callaway I-70 EB 2.0 Rural Freeway Cable Barrier 30 15000 MO08 Callaway I-70 EB 2.0 Rural Freeway Cable Barrier 30 11000 MO09 Montgomery I-70 EB 2.0 Rural Freeway Cable Barrier 30 11000 MO10 Montgomery I-70 EB 2.0 Rural Freeway Cable Barrier 30 15000 MO11 St Charles I-70 EB 2.1 Urban Freeway Concrete Barrier 16 80000 MO12 Clay I-35 SB 3.0 Urban Freeway Concrete Barrier 25 36400 Ohio Sites OH01 Ashland I-71 NB 1.0 Rural Freeway Cable Barrier 40 42000 OH02 Butler I-75 NB 1.0 Urban Freeway Concrete Barrier 38 97000 OH03 Butler I-75 NB 1.0 Urban Freeway Concrete Barrier 34 97000 OH04 Clark I-70 NB 1.0 Rural Freeway Cable Barrier 75 57000 OH05 Clark I-70 NB 1.0 Rural Freeway Cable Barrier 75 57000 OH06 Montgomery I-675 NB 2.1 Urban Freeway Cable Barrier 80 40000 OH07 Montgomery I-75 NB 1.0 Urban Freeway Cable Barrier 150 96000 OH08 Montgomery I-75 NB 1.0 Urban Freeway Cable Barrier 160 104000 OH09 Montgomery I-75 NB 1.0 Urban Freeway Cable Barrier 50 104000 OH10 Montgomery I-75 NB 1.0 Urban Freeway Cable Barrier 50 104000 OH11 Richland I-71 NB 1.0 Rural Freeway Cable Barrier 40 40000 OH12 Richland I-71 NB 1.0 Rural Freeway Cable Barrier 30 39000 OH13 Richland I-71 NB 1.0 Rural Freeway Cable Barrier 30 40000 Washington Sites WA02 Cowlitz I-5 NB 7.2 Rural Freeway Concrete Barrier 30 45000 WA03 King I-5 SB 1.1 Urban Freeway Concrete Barrier 20 229000 WA04 King I-5 NB 0.7 Urban Freeway Concrete Barrier 20 208000 WA05 King I-5 SB 1.4 Urban Freeway Concrete Barrier 16 211000 WA06 Snohomish I-5 NB 2.2 Urban Freeway Concrete Barrier 20 129000 WA07 Skagit I-5 NB 2.3 Urban Freeway Guardrail 42 62000 WA08 Grays Harbor US 12 EB 5.5 Rural Divided Nonfreeway Cable Barrier 30 19000 WA10 King I-90 WB 2.8 Rural Freeway Nontraversable Variable 56000 WA11 King I-90 WB 3.5 Rural Freeway Nontraversable Variable 56000 WA12 King I-90 WB 3.9 Rural Freeway Nontraversable Variable 30000 WA13 King/Kittitas I-90 EB 1.0 Rural Freeway Concrete Barrier 22 30000 WA14 Kittitas I-90 EB 1.3 Rural Freeway Concrete Barrier 24 30000 a The median type and width represents the character of the median for all or most of the length of the site. Table 4-1. Characteristics of field study sites. other vehicle. Single-vehicle loss-of-control crashes included crashes initiated by driver inattention, driver distraction, or driver fatigue, as well as other crashes not related to vehicle interactions (e.g., tire blowouts, loss of control on curves, striking or avoiding an animal, etc.). A total of 73 percent of all median-related crashes were classified as single-vehicle loss-of-control crashes. Vehicle-interaction crashes are those in which it appears that the vehicle would not have left the road and entered the median, but for the interaction of that vehicle with other vehicles. This category includes crashes in which there was a collision on the roadway prior to a vehi- cle entering the median, crashes in which the vehicle that

42 Site number Driver workload level Contributing factors CA01 Low-Moderate On-ramp Long tangent On-ramp located after 12-mi level tangent CA02 Low-Moderate Horizontal curve Truck stopping area on wide shoulder CA04 Low-Moderate Off-ramp On-ramp Bridge CA05 High On-ramp Upgrade Bridge On-ramp on upgrade with short acceleration lane upstream of bridge Low-speed merge due to upgrade CA07 Moderate-High On-ramp Off-ramp Closely spaced on-ramp followed by off-ramp (short weaving area) Closely spaced on-ramps CA10 Low Off-ramp CA11 Moderate-High On-ramp Off-ramp Closely spaced on-ramp followed by off-ramp (short weaving area) Limited sight-distance to on-ramp Mainline lane drop CA13 High On-ramp Tight radius curve on on-ramp Low-speed merge due to on-ramp curve Off ramp Sag vertical curve CA14 Moderate-High On-ramp Off-ramp Bridge Closely spaced on-ramp followed by off-ramp with bridge in short weaving area Tight radius horizontal curve on on-ramp Low-speed merge due to on-ramp curve CA15 Moderate Off-ramp On-ramp Low-speed merge due to short on-ramp MO01 Low Downgrade Bridge Horizontal curve MO02 Moderate Horizontal curve Upgrade Off-ramp On-ramp Crest vertical curve Off-ramp with crest vertical curve that limits sight-distance near start of taper On-ramp on upgrade Low-speed merge due to grade MO03 Moderate Curve Intersections Intersection on curve MO04 Low Horizontal curve Long tangent Horizontal curve after long tangent At-grade intersections MO05 Low-Moderate On-ramp Horizontal curve Downgrade On-ramp on horizontal curve on downgrade Low-speed merge due to mainline curve MO06 Low Horizontal curve Downgrade Horizontal curve on downgrade MO07 Low-Moderate Off-ramp On-ramp High truck volume on on-ramp MO08 Very Low No contributing factors noted Table 4-2. Summary of driver workload levels and contributing factors by site.

43 Site number Driver workload level Contributing factors MO09 Very Low No contributing factors noted MO10 Moderate Horizontal curve Off-ramp On-ramp Trees block sight-distance to on-ramp MO11 High Off-ramp On-ramp Horizontal curve Downgrade On-ramp with multilane entrance (three entering lanes) on slight curve on downgrade (curve has influence even though slight) Closely spaced on-ramps Off-ramp with lane-drop exit Bridge MO12 High Off-ramp On-ramp Left-side on-ramp Closely spaced on-ramps (one left-side, one right-side) Closely spaced on-ramp followed by off-ramp (short weaving area) OH01 Low Bridge OH02 Low No contributing factors noted OH03 Low No contributing factors noted OH04 High On-ramp Off-ramp Off-ramp with short deceleration lane Downgrade Bridge Closely spaced on-ramp followed by off-ramp with bridge between (within full cloverleaf interchange) OH05 Low Bridge OH06 High On-ramp Horizontal curve Major merge area (two freeways merge) Mainline lane drop On-ramp with tight horizontal curve Low-speed merge due to on-ramp curve Closely spaced on-ramps OH07 Moderate Off-ramp On-ramp Bridge Downgrade OH08 Moderate Upgrade Bridge Horizontal curve Crest vertical curve Off-ramp Off-ramp at crest vertical curve on horizontal curve On-ramp OH09 Low No contributing factors noted OH10 Moderate Sharp horizontal curve Off-ramp OH11 Low-Moderate Off-ramp On-ramp Horizontal curve OH12 Moderate Off-ramp Horizontal curve OH13 Moderate-High Bridge On-ramp Downgrade Closely spaced on-ramps WA02 Low-Moderate Off-ramp On-ramp Bridge Table 4-2. (Continued). (continued on next page)

44 Site number Driver workload level Contributing factors WA03 Very High Tunnel HOV lane Off-ramp Closely spaced off-ramps Off-ramp with multilane exit Off-ramp with lane-drop exit Horizontal curve WA04 High Downgrade Horizontal curve On-ramp Off-ramp Downgrade ending in horizontal curve Left-side on-ramp Closely spaced on-ramp followed by off-ramp (one left-side, one right-side) WA05 High On-ramp Off-ramp Left-side off-ramp Off-ramp with lane-drop exit Horizontal curve Downgrade Sag vertical curve On-ramp with merge area in sag vertical curve WA06 High Downgrade On-ramp Off-ramp Off-ramp with multilane exit Off-ramp with lane-drop exit Bridge Closely spaced on-ramp followed by off-ramp with two-lane lane-drop exit WA07 Moderate Narrow bridge Off-ramp Downgrade Off-ramp just beyond narrow bridge on downgrade Off-ramp with multilane exit On-ramp WA08 Low-Moderate At-grade intersections WA10 Moderate Downgrade On-ramp Sharp horizontal curve On-ramp with high truck volume Sharp horizontal curve on downgrade Sag vertical curve Upgrade WA11 Moderate On-ramp Horizontal curve Downgrade Crest vertical curve On-ramp on mainline curve on downgrade Upgrade Horizontal curve at crest vertical curve Long horizontal curve (90o turn) Truck stopping area on wide shoulder WA12 Moderate-High Steep downgrade Horizontal curve Horizontal curve on steep downgrade Site begins downstream of an on-ramp WA13 High Upgrade Downgrade Sharp horizontal curve Off-ramp Off-ramp on horizontal curve on downgrade WA14 High Horizontal curve Off-ramp Reverse curves (four in sequence) Off-ramp on horizontal curve Downgrade Note: Contributing factors with combined effects are noted together. Table 4-2. (Continued).

45 Site number Driver workload level Individual contributing factors CA01 Low-Moderate On-ramp Long tangent CA02 Low-Moderate Horizontal curve Truck stopping area on wide shoulder CA04 Low-Moderate Off-ramp On-ramp Bridge CA05 High On-ramp Upgrade Bridge CA07 Moderate-High On-ramp Off-ramp CA10 Low Off-ramp CA11 Moderate-High On-ramp Off-ramp Mainline lane drop CA13 High On-ramp Off ramp Sag vertical curve CA14 Moderate-High On-ramp Off-ramp Bridge CA15 Moderate Off-ramp On-ramp MO01 Low Downgrade Bridge Horizontal curve MO02 Moderate Horizontal curve Upgrade Off-ramp On-ramp Crest vertical curve MO03 Moderate Horizontal curve At-grade intersections MO04 Low Horizontal curve Long tangent At-grade intersections MO05 Low-Moderate On-ramp Horizontal curve Downgrade MO06 Low Horizontal curve Downgrade MO07 Low-Moderate Off-ramp On-ramp MO08 Very Low No contributing factors noted MO09 Very Low No contributing factors noted MO10 Moderate Horizontal curve Off-ramp On-ramp MO11 High Off-ramp Off-ramp with lane-drop exit On-ramp Horizontal curve Downgrade Bridge MO12 High Off-ramp On-ramp OH01 Low Bridge OH02 Low No contributing factors noted OH03 Low No contributing factors noted OH04 High On-ramp Off-ramp Downgrade Bridge Table 4-3. Summary of driver workload levels and individual contributing factors by site. (continued on next page)

46 Site number Driver workload level Individual contributing factors OH07 Moderate Off-ramp On-ramp Bridge Downgrade OH08 Moderate Upgrade Bridge Horizontal curve Crest vertical curve Off-ramp On-ramp OH09 Low No contributing factors noted OH10 Moderate Sharp horizontal curve Off-ramp OH11 Low-Moderate Off-ramp On-ramp Horizontal curve OH12 Moderate Off-ramp Horizontal curve OH13 Moderate-High Bridge On-ramp Downgrade WA02 Low-Moderate Off-ramp On-ramp Bridge WA03 Very High Tunnel HOV lane Off-ramp Off-ramp with lane-drop exit Horizontal curve WA04 High Downgrade Horizontal curve On-ramp Off-ramp WA05 High On-ramp Off-ramp Off-ramp with lane-drop exit Horizontal curve Downgrade Sag vertical curve WA06 High Downgrade On-ramp Off-ramp Off-ramp with lane-drop exit Bridge WA07 Moderate Narrow bridge Off-ramp Downgrade On-ramp WA08 Low-Moderate At-grade intersections WA10 Moderate Downgrade On-ramp Sharp horizontal curve Sag vertical curve Upgrade WA11 Moderate On-ramp Horizontal curve Downgrade Crest vertical curve Upgrade Truck stopping area on wide shoulder OH05 Low Bridge OH06 High On-ramp Horizontal curve Major merge area (two freeways merge) Mainline lane drop Table 4-3. (Continued).

47 Site number Driver workload level Individual conributing factors WA12 Moderate-High Steep downgrade Horizontal curve Site begins downstream of an on-ramp WA13 High Upgrade Downgrade Sharp horizontal curve Off-ramp WA14 High Horizontal curve Off-ramp Downgrade Note: This table lists all contributing factors whether those factors contribute to high driver-workload individually or in combination with other factors. Table 4-3. (Continued). Site number Driver workload level Combined contributing factors CA01 Low-Moderate On-ramp located after 12-mi level tangent CA02 Low-Moderate No combined contributing factors at this site CA04 Low-Moderate No combined contributing factors at this site CA05 High On-ramp on upgrade with short acceleration lane upstream of bridge Low-speed merge due to upgrade CA07 Moderate-High Closely spaced on-ramp followed by off-ramp (short weaving area) Closely spaced on-ramps CA10 Low No combined contributing factors at this site CA11 Moderate-High Closely spaced on-ramp followed by off-ramp (short weaving area) Limited sight-distance to on-ramp CA13 High Tight radius curve on on-ramp Low-speed merge due to on-ramp curve CA14 Moderate-High Closely spaced on-ramp followed by off-ramp with bridge in short weaving area Tight radius horizontal curve on on-ramp Low-speed merge due to on-ramp curve CA15 Moderate Low-speed merge due to short on-ramp MO01 Low No combined contributing factors at this site MO02 Moderate Off-ramp with crest vertical curve that limits sight-distance near start of taper On-ramp on upgrade Low-speed merge due to grade MO03 Moderate Intersection on horizontal curve MO04 Low Horizontal curve after long tangent MO05 Low-Moderate On-ramp on horizontal curve on downgrade Low-speed merge due to mainline curve MO06 Low Horizontal curve on downgrade MO07 Low-Moderate High truck volume on on-ramp MO08 Very Low No contributing factors noted MO09 Very Low No contributing factors noted MO10 Moderate Trees block sight-distance to on-ramp MO11 High On-ramp with multilane entrance (three entering lanes) on slight curve on downgrade (curve has influence even though slight) Closely spaced on-ramps MO12 High Left-side on-ramp Closely spaced on-ramps (one left-side, one right-side) Closely spaced on-ramp followed by off-ramp (short weaving area) Table 4-4. Summary of driver workload levels and combined contributing factors by site. (continued on next page)

48 entered the median was trying to avoid collision with another vehicle, and other crashes in which vehicle–vehicle inter- actions contributed to the crash. Crashes that involve a vehicle swerving to avoid being cut off by another vehicle changing lanes or swerving to avoid another vehicle stopping suddenly were classified as vehicle-interaction crashes. Thus, vehicle- interaction crashes involve all multiple-vehicle collisions and some single-vehicle crashes that involve avoiding a collision with another vehicle. A total of 27 percent of all median- related crashes were classified as vehicle-interaction crashes. Approximately one-third of vehicle-interaction crashes did not involve a collision with a second vehicle. 4.6.2 Pavement Surface Condition Pavement surface condition was classified based on the pavement surface condition at the time of the crash, as reported by the investigating officer. The category of dry-pavement crashes includes all crashes for which the investigating officer reported dry-pavement conditions at the time of the crash. The category of wet pavement, snow, and ice crashes includes all other pavement surface conditions, consisting primarily of crashes for which the officer explicitly reported wet-pavement conditions at the time of the crash, but also including muddy, snow-, ice-, or slush-covered roadways. In three of the four states, loss of control on wet or snow- covered roads resulting in single-vehicle crashes were the most frequent precursor to median-related crashes, account- ing for 39 percent of median-related crashes in Missouri, 58 percent in Ohio, and 63 percent in Washington. In con- trast, only 16 percent of median-related crashes in California involved single-vehicle loss of control on wet or snow-covered roads. This observed difference likely reflects exposure to wet and snow-covered roads—there is proportionately more dry Site number Driver workload level Combined Contributing factors WA06 High Off-ramp with multilane exit Closely spaced on-ramp followed by off-ramp with two-lane lane-drop exit WA07 Moderate Off-ramp just beyond narrow bridge on downgrade Off-ramp with multilane exit WA08 Low-Moderate No combined contributing factors at this site WA10 Moderate On-ramp with high truck volume Sharp horizontal curve on downgrade WA11 Moderate On-ramp on mainline curve on downgrade Horizontal curve at crest vertical curve Long horizontal curve (90o turn) WA12 Moderate-High Horizontal curve on steep downgrade WA13 High Off-ramp on horizontal curve on downgrade WA14 High Reverse curves (four in sequence) Off-ramp on horizontal curve Note: Contributing factors with combined effects are noted together. OH01 Low Bridge OH02 Low No contributing factors noted OH03 Low No contributing factors noted OH04 High Off-ramp with short deceleration lane Closely spaced on-ramp followed by off-ramp with bridge between (within full cloverleaf interchange) OH05 Low No combined contributing factors at this site OH06 High On-ramp with tight horizontal curve Low-speed merge due to on-ramp curve Closely spaced on-ramps OH07 Moderate No combined contributing factors at this site OH08 Moderate Off-ramp at crest vertical curve on horizontal curve OH09 Low No contributing factors noted OH10 Moderate No combined contributing factors at this site OH11 Low-Moderate No combined contributing factors at this site OH12 Moderate No combined contributing factors at this site OH13 Moderate-High Closely spaced on-ramps WA02 Low-Moderate No combined contributing factors at this site WA03 Very High Closely spaced off-ramps Off-ramp with multilane exit WA04 High Downgrade ending in horizontal curve Left-side on-ramp Closely spaced on-ramp followed by off-ramp (one left-side, one right-side) WA05 High Left-side off-ramp On-ramp with merge area in sag vertical curve Table 4-4. (Continued).

49 Description of contributing factor Total Contributing factors by driver workload level Very low Low Low- moderate Moderate Moderate- high High Very high RAMPS 80 0 4 8 22 9 34 3 On Ramps 44 0 2 5 14 5 18 0 On ramp 22 2 3 9 1 7 Closely spaced on ramps 5 2 3 Left-side on-ramp 2 2 Multilane entrance 1 1 Merge area in sag vertical curve 1 1 Downstream of on-ramp 1 1 High truck volume on on-ramp 2 1 1 Major merge 1 1 Short acceleration lane 1 1 Low-speed merge due to curve on ramp 3 1 2 Low-speed merge due to mainline curve 1 1 Low-speed merge due to short ramp 1 1 Low-speed merge due to grade 1 1 Limited sight-distance to on ramp 2 2 Off Ramps 29 0 2 3 8 1 12 3 Off ramps 20 2 3 7 8 Closely spaced off ramps 1 1 Multilane exit 2 1 1 Off-ramp with lane drop 3 2 1 Left-side off ramp 1 1 Short deceleration lane 1 1 Limited sight-distance to off ramp 1 1 On- and Off-Ramps 7 0 0 0 0 3 4 0 Closely spaced on- and off-ramp 7 3 4 HORIZONTAL CURVES 25 0 4 4 10 1 5 1 Horizontal curves 17 4 4 5 1 2 1 Sharp horizontal curve 3 2 1 Long horizontal curve 1 1 Sequence of reverse curves 1 1 Downgrade ending in curve 1 1 Crest vertical curve on horizontal curve 2 2 VERTICAL ALIGNMENT 21 0 2 1 8 2 8 0 Downgrade 13 2 1 4 1 5 Steep downgrade 1 1 Upgrade 5 3 2 Sag vertical curve 2 1 1 OTHER 24 0 5 3 7 2 5 2 Bridge 12 4 1 2 1 4 Narrow bridge 1 1 At-grade intersection 3 1 1 1 At-grade intersection on curve 1 1 Mainline lane drop 2 1 1 Tunnel 1 1 HOV lane 1 1 Widened shldr/truck stopping area 3 1 2 Total Observations 150 0 15 16 47 14 52 6 Percentage of Observations 0.0 10.0 10.7 31.3 9.3 34.7 4.0 Table 4-5. Frequency distribution of individual and combined contributing factors.

50 weather in California in contrast to the other three states, and much more snow, ice, and slush in Ohio and Washington than in Missouri. Table 4-8 compares the mean number of days with precipi- tation for California, Missouri, Washington, and Ohio based on data from the U.S. National Oceanic and Atmospheric Administration (NOAA) National Climatic Data Center. Days of precipitation encompass days with 0.01 inch of precipitation or more. The averages provided in the table were calculated from the mean number of days of precipitation from a range of cities across the state to provide a representation of the various weather conditions throughout the region. The following cities were used for deriving the averages for each state: • California: Bishop, Los Angeles, Santa Barbara, Sacramento, San Francisco, Blue Canyon, Mount Shasta, and Eureka; • Ohio: Dayton, Toledo, Columbus, Akron, Cleveland, and Youngstown; • Missouri: Kansas City, St. Louis, and Columbia; and • Washington: Yakima, Spokane, Seattle. The number of years over which the weather data have been collected ranges from 25 to 101 years prior to 2011. As might be expected, the greater the percent of days per year with precipitation, the greater the percent of median crashes that were on wet and snow-covered roads. It should be noted though that the percent of days per year with 0.01 inch of precipitation or more ranged from 16 per- cent to 39 percent. On those days with precipitation, some of the time the road would be dry. In other words, considering days with no precipitation as well as there being only part of the day with wet or snow-covered road conditions, the major- ity of the time roads would have been dry. However, combin- ing the results for all four states, there were more crashes on wet and snow-covered roads (average 44 percent) than would be suggested by a very conservative estimate of exposure to wet and snow-covered road conditions—days per year with at least 0.01 inch of precipitation (average 30 percent). More than twice as many single-vehicle loss-of-control median-related crashes occurred on wet and snow-covered roads (49 percent) than on dry roads (23 percent), despite dry- road conditions being present at least 2.3 times (70 percent vs. 30 percent) more often than wet and snow-covered roads. Thus, wet and snow-covered roads are overinvolved in single- vehicle loss-of-control median-related crashes by a factor of at least 4.6, given that days with 0.01 inch of precipitation likely have dry hours. This finding suggests that wet and snow-covered roads are a major risk factor for single-vehicle loss-of-control median- related crashes. A meta-analysis based on 34 published studies reports average increases in crash rates of 71 percent for rain and 84 percent for snow (52). The data reviewed here would suggest a much higher risk for single-vehicle loss-of-control median-related crashes as compared to other crash types. A total of 19 percent of all median-related crashes were vehicle-interaction crashes that occurred on dry roads, while 9 percent were vehicle-interaction crashes on wet and snow- covered roads. In other words, vehicle-interaction crashes Description of contributing factor Total Contributing factors by driver workload level Very low Low Low- moderate Moderate Moderate- high High Very high On-ramp 28 6 8 5 9 Off-ramp 26 1 4 8 3 9 1 Horizontal curve 19 3 3 6 1 5 1 Sharp horizontal curve 3 2 1 Bridge 13 3 2 2 2 4 Narrow bridge 1 1 Downgrade 15 2 1 4 1 7 Steep downgrade 1 1 Upgrade 6 4 2 Crest vertical curve 3 3 Sag vertical curve 3 1 2 Long tangent 2 1 1 Mainline lane drop 2 1 1 Lane drop at exit 4 3 1 At-grade intersections 3 1 1 1 Major merge (two freeways merge) 1 1 Tunnel 1 1 HOV lane 1 1 Truck stopping area on wide shoulder 2 1 1 No contributing factors noted 5 2 3 Table 4-6. Frequency distribution of contributing factors by major categories.

51 Site number Number of crashes in a 5-year period by crash type and pavement surface condition Length (mi) Directional AADT (veh/day) Crash rate Single-vehicle loss of control Vehicle interaction Total Dry roads Wet roadsa Dry roads Wet roadsa per mile per year per MVMT California Sites CA01 8 0 5 0 13 2.0 39000 1.30 0.091 CA02 10 0 4 0 14 2.0 39000 1.40 0.098 CA04 5 4 4 0 13 2.0 58000 1.30 0.061 CA05 8 0 9 0 17 1.0 61000 3.40 0.153 CA07 1 5 3 0 9 2.0 61000 0.90 0.040 CA10 7 0 4 0 11 1.0 75000 2.20 0.080 CA11 2 1 5 1 9 1.0 72000 1.80 0.068 CA13 6 10 6 4 26 1.0 67000 5.20 0.213 CA14 5 1 4 1 11 1.0 67000 2.20 0.090 CA15 4 1 10 1 16 1.0 64000 3.20 0.137 Missouri Sites MO01 4 8 3 4 19 2.0 15000 1.90 0.347 MO02 14 27 14 11 66 3.0 17000 4.40 0.709 MO03 9 1 3 0 13 2.0 7500 1.30 0.475 MO04 4 0 3 0 7 3.6 7000 0.39 0.152 MO05 16 25 11 4 56 10.7 20000 1.05 0.143 MO06 5 9 3 0 17 2.0 17000 1.70 0.274 MO07 10 8 10 3 31 2.0 15000 3.10 0.566 MO08 16 2 7 0 25 2.0 11000 2.50 0.623 MO09 6 3 4 1 14 2.0 11000 1.40 0.349 MO10 7 1 8 0 16 2.0 15000 1.60 0.292 MO11 8 66 10 15 99 2.1 80000 9.43 0.323 MO12 13 2 7 0 22 3.0 36400 1.47 0.110 Ohio Sites OH01 3 7 1 0 11 1.0 42000 2.20 0.144 OH02 2 16 10 7 35 1.0 97000 7.00 0.198 OH03 0 6 3 4 13 1.0 97000 2.60 0.073 OH04 5 10 5 0 20 1.0 57000 4.00 0.192 OH05 1 8 3 2 14 1.0 57000 2.80 0.135 OH06 3 10 3 4 20 2.1 40000 1.90 0.130 OH07 4 9 1 12 26 1.0 96000 5.20 0.148 OH08 4 18 2 2 26 1.0 104000 5.20 0.137 OH09 5 9 3 3 20 1.0 104000 4.00 0.105 OH10 12 10 4 1 27 1.0 104000 5.40 0.142 OH11 8 17 3 3 31 1.0 40000 6.20 0.425 OH12 5 39 4 0 48 1.0 39000 9.60 0.674 OH13 4 27 1 1 33 1.0 40000 6.60 0.452 Washington Sites WA02 49 7 0 4 60 7.2 45000 1.68 0.102 WA03 14 9 3 11 37 1.1 229000 6.67 0.080 WA04 9 6 6 5 26 0.7 208000 7.54 0.099 WA05 21 6 3 11 41 1.5 211000 5.66 0.073 WA06 3 5 8 5 21 2.2 129000 1.89 0.040 WA07 4 3 8 1 16 2.3 62000 1.42 0.063 WA08 5 10 0 1 16 5.5 19000 0.58 0.084 WA10 9 12 2 2 25 2.8 56000 1.77 0.086 WA11 2 21 1 1 25 3.5 56000 1.45 0.071 WA12 9 41 2 2 54 3.9 30000 2.78 0.254 WA13 2 22 1 2 27 1.0 30000 5.19 0.474 WA14 7 55 1 2 65 1.3 30000 9.85 0.899 Total 358 557 215 131 1261 97.4 Percent of Total 28% 45% 17% 10% a Includes all crashes for which the pavement surface condition is reported as other than dry, including wet, muddy, snow-, ice-, or slush-covered roads. Table 4-7. Crash frequency by site for median-related crashes.

52 were only half as likely to occur on wet and snow-covered roads as on dry roads. If it was assumed that road condi- tion was affected the entire day on days in which 0.01 inch of precipitation or more was recorded, one would expect about 30 percent of the vehicle-interaction crashes to occur on wet and snow-covered roads. The fact that approximately 32 percent occur in such conditions suggests that wet and snow-covered roads raise the risk of a vehicle-interaction crash much less than is the case for single-vehicle loss-of- control crashes. 4.6.3 Driver Contributing Factors as Classified by Investigating Officers Much of the analysis performed was based on contrib- uting factors to median-related crashes, as classified by the interdisciplinary team based on review of the crash reports and the field sites. Consideration also was given to the driver contributing factors as classified on the crash reports by the investigating officers. Speed-Related Crashes—The most frequent driver con- tributing factor for single-vehicle loss-of-control median- related crashes under wet and snow-covered road conditions, as noted by the investigating officers in the crash reports, was speed (91 percent of single-vehicle loss-of-control crashes on wet and snow-covered roads in Missouri, 71 percent in Ohio, 84 percent in Washington). In contrast, speed was much less frequently cited as a factor in single-vehicle loss-of-control crashes involving dry roads (26 percent of single-vehicle loss- of-control crashes on dry roads in Missouri, 34 percent in Ohio, 30 percent in Washington). With respect to vehicle-interaction crashes on wet and snow-covered roads, speed was less frequently cited than in single-vehicle loss-of-control crashes on wet and snow- covered roads (62 percent of vehicle-interaction crashes on wet and snow-covered roads in Missouri, 16 percent in Ohio, 54 percent in Washington). In vehicle-interaction crashes on dry as compared to wet and snow-covered roads, speed was even less frequently cited (Missouri 12 percent vs. 62 per- cent, Ohio 2 percent vs. 16 percent, Washington 21 percent vs. 54 percent). Given that the investigating officer completing a crash report is generally not in a position to determine what the speed was or how it compared to surrounding vehicles, the attribution of speed as the cause of the crash is generally an assumption rather than being based on evidence. Nevertheless, research indicates that while drivers slow down, they do so less than is appropriate given the loss of friction and reduction in visibility that makes vehicle handling more difficult (53). Thus, speed is generally likely to contribute more to wet and snow-covered road crashes, but how large a factor it is cannot really be deter- mined from crash reports. This is particularly evident from the very different findings from Ohio as compared to Missouri and Washington with respect to the percent of various types of crashes involving speed (e.g., Missouri, 62 percent vs. Wash- ington, 54 percent vs. Ohio, 16 percent of vehicle-interaction crashes on wet and snow-covered roads). Inattention/Distraction-Related Crashes—After speed, inattention was the most frequently cited driver contribut- ing factor in median-related crashes. Inattention was much more likely to be cited on dry than on wet and snow-covered roads (12 percent of dry-road crashes vs. 2 percent of wet and snow-covered road crashes), and far more likely to be cited in Missouri than Ohio (10 percent vs. 1 percent of all crashes). Crashes with No Contributing Factor Cited—Overall investigating officers in Ohio and Missouri were less likely to cite a driver contributing factor than police in Washing- ton. This finding suggests differences in police culture with respect to assigning contributing factors that may bias find- ings. Given how little evidence police have to depend on to make attributions of speed, inattention, etc., some subjectiv- ity and reliance on assumptions is to be expected. Understanding and utilizing all of the collected data allowed the team to classify each of the field investigation sites by work- load level. Table 4-9 presents the site-by-site crash frequency data grouped into the seven driver workload categories. 4.6.4 Cross-Median Collisions The 1,261 crashes shown in Table 4-9 each involved a median encroachment. Most of those crashes involved a vehi- State Average number of days per year with precipitation Percentage of days per year with precipitation Percentage of median crashes that occurred on wet and snow- covered roads California 57.5 16% 16% Missouri 103.0 28% 39% Ohio 140.6 39% 58% Washington 136.4 37% 63% Table 4-8. Differences in precipitation frequency by state compared to proportion of crashes on wet and snow-covered roads.

Site number Driver workload Number of crashes in a 5-year period by crash type and pavement surface condition Length (mi) Directional AADT (veh/day) Crash rateLoss of control Vehicle interaction TotalDry roads Wet roadsa Dry roads Wet roadsa per mile per year per MVMT Very low driver workload MO08 Very low 16 2 7 0 25 2.0 11000 2.50 0.623 MO09 Very low 6 3 4 1 14 2.0 11000 1.40 0.349 Subtotal Very low 22 5 11 1 39 4.0 1.95 0.486 56% 13% 28% 3% Low driver workload CA10 Low 7 0 4 0 11 1.0 75000 2.20 0.080 MO01 Low 4 8 3 4 19 2.0 15000 1.90 0.347 MO04 Low 4 0 3 0 7 3.6 7000 0.39 0.152 MO06 Low 5 9 3 0 17 2.0 17000 1.70 0.274 OH01 Low 3 7 1 0 11 1.0 42000 2.20 0.144 OH02 Low 2 16 10 7 35 1.0 97000 7.00 0.198 OH03 Low 0 6 3 4 13 1.0 97000 2.60 0.073 OH05 Low 1 8 3 2 14 1.0 57000 2.80 0.135 OH09 Low 5 9 3 3 20 1.0 104000 4.00 0.105 Subtotal Low 31 63 33 20 147 13.6 2.16 0.144 21% 43% 22% 14% Low-moderate driver workload CA01 Low-moderate 8 0 5 0 13 2.0 39000 1.30 0.091 CA02 Low-moderate 10 0 4 0 14 2.0 39000 1.40 0.098 CA04 Low-moderate 5 4 4 0 13 2.0 58000 1.30 0.061 MO05 Low-moderate 16 25 11 4 56 10.7 20000 1.05 0.143 MO07 Low-moderate 10 8 10 3 31 2.0 15000 3.10 0.566 OH11 Low-moderate 8 17 3 3 31 1.0 40000 6.20 0.425 WA08 Low-moderate 5 10 0 1 16 5.5 19000 0.58 0.084 Subtotal Low-moderate 62 64 37 11 174 25.2 1.38 0.144 36% 37% 21% 6% Moderate driver workload CA15 Moderate 4 1 10 1 16 1.0 64000 3.20 0.137 MO02 Moderate 14 27 14 11 66 3.0 17000 4.40 0.709 MO03 Moderate 9 1 3 0 13 2.0 7500 1.30 0.475 MO10 Moderate 7 1 8 0 16 2.0 15000 1.60 0.292 OH07 Moderate 4 9 1 12 26 1.0 96000 5.20 0.148 OH08 Moderate 4 18 2 2 26 1.0 104000 5.20 0.137 OH10 Moderate 12 10 4 1 27 1.0 104000 5.40 0.142 OH12 Moderate 5 39 4 0 48 1.0 39000 9.60 0.674 WA02 Moderate 49 7 0 4 60 7.2 45000 1.68 0.102 WA07 Moderate 4 3 8 1 16 2.3 62000 1.42 0.063 WA10 Moderate 9 12 2 2 25 2.8 56000 1.77 0.086 WA11 Moderate 2 21 1 1 25 3.5 56000 1.45 0.071 Subtotal Moderate 123 149 57 35 364 28 2.63 0.151 34% 41% 16% 10% Table 4-9. Crash frequency by site for median-related crashes for specific driver workload levels. (continued on next page)

Site number Driver workload Number of crashes in a 5-year period by crash type and pavement surface condition Length (mi) Directional AADT (veh/day) Crash rateLoss of control Vehicle interaction TotalDry roads Wet roadsa Dry roads Wet roadsa per mile per year per MVMT Moderate-high driver workload CA07 Moderate- high 1 5 3 0 9 2.0 61000 0.90 0.040 CA11 Moderate- high 2 1 5 1 9 1.0 72000 1.80 0.068 CA14 Moderate- high 5 1 4 1 11 1.0 67000 2.20 0.090 OH13 Moderate-high 4 27 1 1 33 1.0 40000 6.60 0.452 WA06 Moderate-high 3 5 8 5 21 2.2 129000 1.89 0.040 WA12 Moderate-high 9 41 2 2 54 3.9 30000 2.78 0.254 Subtotal Moderate-high 24 80 23 10 137 11.1 2.47 0.107 18% 58% 17% 7% High driver workload CA05 High 8 0 9 0 17 1.0 61000 3.40 0.153 CA13 High 6 10 6 4 26 1.0 67000 5.20 0.213 MO11 High 8 66 10 15 99 2.1 80000 9.43 0.323 MO12 High 13 2 7 0 22 3.0 36400 1.47 0.110 OH04 High 5 10 5 0 20 1.0 57000 4.00 0.192 OH06 High 3 10 3 4 20 2.1 40000 1.90 0.130 WA04 High 9 6 6 5 26 0.7 208000 7.54 0.099 WA05 High 21 6 3 11 41 1.5 211000 5.66 0.073 WA13 High 2 22 1 2 27 1.0 30000 5.19 0.474 WA14 High 7 55 1 2 65 1.3 30000 9.85 0.899 Subtotal High 82 187 51 43 363 14.7 4.94 0.187 23% 52% 14% 12% Very high driver workload WA03 Very high 14 9 3 11 37 1.1 229000 6.67 0.080 Subtotal Very high 14 9 3 11 37 1.1 6.67 0.080 38% 24% 8% 30% Total 358 557 215 131 1261 97.4 Percent of Total 28% 45% 17% 10% aIncludes all crashes for which the pavement surface condition is reported as other than dry, including wet, muddy, snow-, ice-, or slush-covered roads. Table 4-9. (Continued).

55 cle striking an object, overturning, or coming to rest before crossing the median. Indeed, only 12 percent of the 97.4 miles of interdisciplinary field review sites had traversable medians where it is physically possible to cross the median without encountering a median barrier, fixed objects, or nontravers- able terrain. During the 5-year study period, the 47 interdisciplin- ary field review sites experienced only two CMC crashes in which a vehicle crossed the entire median, entered the opposing roadway, and collided with an opposing direc- tion vehicle. In addition, there were three NCMC crashes in which a vehicle crossed the entire median and entered the opposing roadway, but did not collide with an oppos- ing direction vehicle. It should be noted that, at the one site that experienced both a CMC and an NCMC crash, the state highway agency has since installed a cable median barrier. 4.7 Summary of Contributing Factors Based on the interdisciplinary field reviews and crash reviews, the following contributing factors appear most involved in determining driver workload and potentially contributing to the initiation of median-related crashes: • On-ramps (especially closely spaced on-ramps), on-ramps with low-speed merges, on-ramps with high entering truck volume, left-side on-ramps, and on-ramps with limited sight-distance to the mainline; • Off-ramps, especially off-ramps with lane drops and multi- lane exits; • Closely spaced on- and off-ramps; • Sharp horizontal curves; • Steep grades, especially steep downgrades, but also includ- ing steep upgrades; and • Wet or snow-covered road conditions.