U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Safety

FHWA Home / Safety / Roadway Departure / Roadway Visibility Research Needs Assessment

Roadway Visibility Research Needs Assessment

 < Previous Table of Content Next >

5. Current Gaps in Practice and Research

As part of this study, researchers identified gaps in the current body of knowledge regarding the three areas of roadway departure, intersection, and pedestrian/bicycle safety. For each of these topics, an initial problem statement was developed and is presented in Appendix C.

Roadway Departure Safety

Placement of Roadway Lighting

Determining where lighting is most effective at reducing crashes could inform policy on roadway lighting design. Data made available in existing databases like SHRP2 can be a target of further research.

Geometric Considerations

Roadway departures occur on both tangents and curves. Determining the effectiveness of lighting a curve versus a tangent can be informative regarding how lighting can best be implemented. Current standards for curved roadways include instruction for pole placement to reduce the likelihood of a collision by an errant vehicle, but less research pertains to what areas of a curved roadway should be lighted to reduce departures.

Lighting and Inclement Weather

Adverse weather can reduce visibility of the roadway and pavement markings. The impact of lighting in adverse weather on many different road types and in combination with varying pavement markings and sign materials has yet to be fully explored. Policies rarely consider the diminished visibility caused by adverse weather conditions.

Determining When to Use Roadway Lighting

The use of roadway lighting is often determined by levels of annual average daily traffic in combination with areas of conflict, municipal regulations on lighting, economic viability, functional class, and zoning. Most traffic, however, occurs during the daytime. If roadway lighting is partially determined by traffic volumes, a count of nighttime traffic should be used as a decision threshold because it directly matches the traffic (vehicle miles traveled [VMT] for RwD, entering traffic for intersections, and pedestrians/bicyclists for that focus area) that would use the lighting. Another determining factor is public opinion on whether an area does or does not warrant lighting.

LED technology is slowly replacing HPS lighting conventionally used in most roadway lighting applications. LEDs can be controlled to output varying intensities and color, and achieve particular angles to satisfy ordinances. This adaptable aspect of an LED can foster more research in these specific areas that has not been possible with HPS and other light sources.

Vehicle Lighting

Headlamps are to have a minimum height of 22 inches, with a maximum of 54 inches. Taillamps are to be a minimum height of 15 inches, and no higher than 72 inches.(32) These measurements are to be made when there are no occupants or baggage in the vehicle. The additional weight would reduce the height of these lights. Guidelines for the design of vertical curvature and passing sections are based on stopping and passing sight distances that usually assume vehicle headlamps and taillamps are at a height of 24 inches.(7) With the allowable taillamp height lower than the assumption commonly used in design (and possibly even lower when the weight of passengers and baggage is accounted for), there are potential safety concerns related to stopping sight distances involving low vehicles on crest vertical curves.

It is understood that the characteristics of all vehicles cannot be accounted for in highway design, but there is a disconnect between the minimum specification for the heights of vehicle lamps and the values commonly used in design. Another issue related to headlamps and vertical curvature is the shadow zone that occurs beyond the crest of a vertical curve, where an object in the road may not be visible until it is within the direct line of an approaching vehicle’s headlamps. The AASHTO Green Book suggests that with the assumed 24-inch height of headlamps, an object 16 inches above the roadway will be within the line of the headlamps at a distance equal to stopping sight distance.(7) This means drivers will typically see objects 16 inches above the pavement in time to stop. However, this may not be true for objects less than 16 inches high or if the headlamps are less than 24 inches high (which is currently permissible). Research that leads to coordination between the headlamp specifications and typical design assumptions is recommended.

The advent of LED headlamps has affected how drivers perceive and react to oncoming headlamps and glare. Research indicates that LED headlamp sources are better than halogen headlamps and comparable to high-intensity discharge headlamps in terms of photometric performance and visibility. The blue-white color produced by the headlamps does reportedly cause glare and discomfort to oncoming drivers.(33) Research into alternative LED colors and more precise beam patterns may mitigate the discomfort to other drivers and maintain photometric and visibility properties. Additionally, the colors produced by the different types of headlamps may have unique effects on sign visibility that are yet to be identified.

Vehicle headlamps are often overlooked when developing designs and policies for roadway lighting and signing. There are several positive safety effects of new headlamp technologies, such as adaptive headlamps that reduce glare for oncoming vehicles or turn in the direction of curves. However, these and other changes in headlamp illumination may have detrimental effects on visibility that should be considered in future research and policies. A new report by the Insurance Institute for Highway Safety (IIHS) demonstrates that most headlights need improvement.(34)

Standardized Police Reporting

The implementation of lighting is partially driven by crash data obtained from on-scene crash reporting. Currently, there are no known standards among law enforcement agencies for reporting the placement and contribution of lighting for crashes. An effort to collect data on how different agencies interpret and apply knowledge related to lighting for crashes could improve how crash data are reported and be a step toward standardizing the reporting process. Currently the Model Minimum Uniform Crash Criteria (MMUCC) includes a field dark—lighted.

Definition of Sight Distance

Sight distance is defined by AASHTO as the length of roadway visible to the driver. It can be categorized into horizontal and vertical sight distance: horizontal sight distance is limited by objects and the changing alignment in the horizontal plane, while vertical sight distance is limited by objects and the changing alignment in the vertical plane. Sight distance is generally not specified as being applicable to a certain time of day (daytime or nighttime). However, many of the assumptions used to calculate different sight distances (e.g., stopping sight distance, decision sight distance, passing sight distance) are based on values that may only be relevant for daytime conditions when daylight illuminates objects and the roadway environment.(7)

When discussing the topic of sight distance, the traditional design manuals have little mention of specific values or considerations for nighttime conditions. AASHTO guidelines for stopping sight distance (SSD), for example, are based on the provision of adequate sight distance at every location on a roadway such that drivers traveling at or near the design speed can see a stationary object in the road and stop before reaching it. The AASHTO assumptions for SSD use a driver eye height of 3.5 ft and an object height of 2 ft. (The 2-ft object height is one example where nighttime conditions are considered because most taillamps are 2 ft or higher from the ground. However, there is still a vulnerability from encounters with shorter objects or vehicles lower to the ground, as mentioned above.) There are no specifics to the shape, color, contrast, or reflection of the object, implying that this is a daytime design element. In general, decision sight distance uses the same basic models as SSD but with longer perception-reaction times, consequently neglecting the differences between day and night conditions.(7)

Pavement Marking Retroreflectivity

There is still some debate about the trade-off between pavement marking retroreflectivity levels and size of pavement markings (width). More research is needed to determine if wider longitudinal pavement markings (i.e., 6 inches) may provide more visibility to drivers and if that gain is enough to provide any relief to the proposed minimum retroreflectivity levels being pursued by FHWA. Also, there is a series of studies available in the last decade showing a consistently stronger link between pavement marking retroreflectivity and safety. However, there is not a general consensus on the relationship, and more work is needed to establish the link. Finally, since machine vision systems are becoming more common, there is a need to determine the pavement marking performance needs for machine drivers (compared to human drivers).

Research is also needed in the area of pavement markings specifically designed to maintain their retroreflective performance conditions under wet nighttime conditions. The crash data reviewed in this report indicate that roadway departure crashes under the category of the dark, not lighted condition were more prevalent in wet conditions compared to clear conditions. This finding seems to indicate the importance of retroreflective performance in wet conditions in terms of roadway departure crashes. However, more research is needed in this area to develop a better understanding of how increased wet nighttime retroreflectivity levels can mitigate roadway departure crashes.

Pavement Markings for Machine Vision

There has and continues to be a fast-paced advancement in vehicle technologies that provide driver assistance such as lane departure warning, lane keep assistance, and automated steering. The primary sensor used today for these systems is a camera mounted in the windshield. These systems are meant to keep vehicles on the road and in their lane. They rely on the ability to detect edge lines and lane lines. Research is currently underway to better understand the performance needs of pavement markings so that the machine vision systems can reliably perform their intended driver assist functions. NCHRP is sponsoring the research and the objective of the research is to develop information on the performance characteristics of pavement markings that affect the ability of machine vision systems to recognize them (http://apps.trb.org/cmsfeed/TRBNetProjectDisplay.asp?ProjectID=4004).

Traffic Sign Visibility

Recent research identifies conditions for which retroreflective traffic signs may be too bright.(35) While the MUTCD identifies minimum levels of maintained retroreflectivity, there are no guidelines specifying when the luminance from a sign causes glare that may be a safety hazard. The glare from signs reduces the distance at which drivers can see hazardous objects and also reduces response times.(36) The concern for glare from signs should extend to the use of digital signs as well. Digital signs can be loosely defined as LED-enhanced signs such as stop signs with flashing LEDs in the border, chevrons with flashing LEDs, or even full-color digital changeable message signs.

The MUTCD states that the brightness of changeable message signs should be adjusted under varying light conditions to maintain legibility, implying a concern that they need to be bright enough to be read during the daytime.(37) However, there is no mention of the issue that changeable message signs may be a source of glare at night. This concern should be especially noted for full-matrix LED signs capable of very high light output. While previous sign research emphasized the need for legibility, the improvements in sign materials and the increasing use of digital infrastructure open up the possibility that these signs are too bright for driver comfort and safety. Future research and applications should address these concerns.

Additionally, there is a need to identify stronger relationships between the retroreflectivity levels of traffic signs and roadway departure crashes. Research has shown some evidence that there appears to be a maximum level of retroreflectivity depending on specific conditions,(35) and FHWA has established minimum levels based on the needs of nighttime drivers in the MUTCD. However, there are no established correlations between retroreflectivity and crashes. In some ways, having recommended guidelines for appropriate sign sheeting selections based on the specific conditions (e.g., roadway, traffic, roadside) would be an alternative way to address this, but again, these would need to be developed.

Speed Limits

Most speed limits are set by legislative action or studies that determine the 85th percentile speed under free-flow conditions. When determined based on speed studies, most often the speed data are collected during the day, meaning that nighttime conditions are not accounted for in selecting speed limits. While separate speed limits are permitted at night, few agencies use nighttime speed limits, and there is no guidance provided in the MUTCD for establishing nighttime speed limits. In addition, there is the concern of over-driving headlamps at night. While no recent study has been completed on this topic, there is a growing concern that the new headlamp trends may be generating a need to take a fresh look at this somewhat controversial topic. On-going research may address this need (NCHRP 17-76, Guidance for Setting Speed Limits).

Horizontal Curve Design and Curve Advisory Speeds

The design of horizontal curves is based on comfortable lateral acceleration rates derived during daytime studies. At night, however, reduced visibility may affect a driver’s threshold level of lateral acceleration where discomfort begins, thus affecting the comfortable curve advisory speed for the curve. While a separate advisory speed for night seems impractical, there may be implications related to safety if drivers are actually not comfortable adopting the advisory speed at night. In an exploratory analysis of unfamiliar driver data collected during both daytime and nighttime conditions, it was found that drivers on curves accept lower levels of lateral acceleration at night than during the day. However, contrasting the reduced lateral acceleration was a finding that driver speeds were higher at night (possibly from cutting curves with a wider path).(18) Further analysis of the implications of curve advisory speeds and nighttime driving could lead to improvements in how these advisory speeds are set and used.

Intersection Safety

Lighting at Rural Intersections

Rural intersections vary in a number of ways, as do the policies and standards for lighting them. The factors involved in fatal crashes at many high-speed intersections in rural areas, whether signalized or not signalized, need to be explored to better inform policy for lighting.

Transient adaptation occurs when drivers travel to and from lit and unlit areas and may cause issues with vision for drivers as they pass through a lighted rural intersection. The general policy is to provide a tapering of lighting so that the transition between the two areas is gradual and comfortable. Exploring the limits of transient adaptation can better inform policies on rural intersection lighting as well as tunnel lighting.

As mentioned, another consideration for rural intersections are those of availability of electrical power, impact on flora and fauna as well as dark sky considerations. Reducing the light to a level that provides the safety benefits without over lighting is critical in these kinds of environments. Using minimum lighting levels reduces the negative impacts of lighting, reduces consumption and provides the opportunity to investigate alternative power sources such as solar and wind generation. Determining this minimum level is another critical component of the required research.

Intersection Sight Distance and Gap Acceptance

A recent NCHRP project has identified the safety impacts of providing intersection sight distance (ISD), showing that fewer crashes occur as ISD increases.(38) Even though ISD is measured based on geometry, there may be issues related to nighttime visibility if the clear sight triangles needed for adequate ISD can be different during the night than during the day. Controls for ISD are based on principles of gap acceptance, which has been derived from daytime observations. A driver’s gap acceptance is a function of perception-reaction time, the time (and distance) used to make an appropriate maneuver, the speed of the conflicting vehicles, and any buffer added for personal preference and safety. The duration of a gap that drivers select may vary with time of day. At night, this may be impacted more because visibility is limited and there is potential for misjudging the distance of a conflicting vehicle. It is possible that what may be an acceptable sight distance during the day may not be adequate at night. Future research may address how gap acceptance changes by time of day, whether there is a need to consider how nighttime gap acceptance may impact ISD and intersection safety, and whether or not intersection lighting would mitigate these effects.

Selecting Phasing for Left Turns

The second edition of the Signal Timing Manual (NCHRP Report 813) provides national guidance on the selection of left-turn phasing modes at signalized intersections. Practitioners select protected or permissive phasing based on criteria such as the number of left and through lanes, sight distance, turning volumes, speeds, and crash history.(39) There is no specific consideration for nighttime conditions. These guidelines can be improved by including special circumstances for nighttime conditions (for example, appropriate thresholds for nighttime crashes rather than total crashes only or turning volumes at night rather than peak-hour volumes). This is a classic example of where reviews of national policies is needed with a focus on nighttime considerations. Review teams would include a nighttime visibility expert along with a team of relevant subject matter experts. A review should also include more than just left phasing, for instance, right urn phasing, right turn on red, and protected/permission phasing.

Skewed Intersections: Visibility of Pedestrians or Bicycles at Night

Intersections with large skew angles are potential safety concerns at night for pedestrians and bicycles approaching the intersection from a conflicting direction. Because a vehicle’s headlamps are directed forward, with only a small amount of illumination distributed away from the center, there may be conditions where drivers are unlikely to see pedestrians or bicycles from some directions, depending on the use of lighting. At night, the visual attention of drivers tends to be concentrated in a smaller area than during the day, meaning that drivers are additionally less likely to search for those approaching pedestrians and bicycles. This concern can be addressed by evaluating the illuminance patterns of headlamps and whether or not the light distributed horizontally at wide angles is enough for a driver to see pedestrians and bicycles.

Signal Warrants

Most of the warrants for determining whether a signal should be considered at an intersection focus on conditions experienced during the daytime. The heaviest traffic in most places tends to occur during morning and afternoon/evening peak hours, so it is reasonable that the listed warrants encourage the engineer to focus on those times of the day. However, special nighttime conditions may exist that warrant the use of a signal at an intersection. Examples may include reduced visibility or an unexpected peak in traffic due to a particular traffic generator. In fact, visibility appears to not be a concern in any signal warrant except indirectly as part of the crash warrant.

Alternative Interchanges and Intersections

Innovations in interchange and intersection design can promote large increases in efficiency, with some intersections (such as roundabouts) providing for the free flow of vehicles. Diverging diamond interchanges and displaced left-turn or continuous-flow intersections are recent concepts specifically designed to address conflicts with turning vehicles. The proper use of lighting and signage at these types of intersections is critical because they are very different from typical stop-controlled or signalized intersections. Pedestrian safety is also a critical component of determining the lighting and signage needs..

Traffic Signs

There is a need to establish recommendations for traffic sign retroreflectivity (materials) and possibly pavement markings. Depending on the type of environment and intersection control (rural versus urban, or stop sign versus signal), there could be justification for different performance levels. In fact, there are no national recommendations or guidelines that a practitioner can use to select the appropriate retroreflective product or performance level for a specific scenario. This is particularly needed for signing at intersections but it is also a general need that extends beyond intersections.

Automation and Controls

With the advent of solid state lighting, controlling where the lighting can be turned on and off as well as dimmed is becoming more important. This technology provides the opportunity to adjust lighting levels and to react to the presence of vehicles and pedestrians. The investigation of the possible options of this type of technology is general for all lighting in the roadway.

Pedestrian and Bicycle Safety

Lighting for Pedestrians

Lighting is commonly designed for vehicles and not for pedestrians. Providing a standard of lighting that meets the criteria for both is a key interest. Lighting for pedestrians is believed to be more crucial in providing the greatest safety benefit at conflict points. Gap acceptances for pedestrians and the perception pedestrians have of how visible they are to drivers are research topics that have been explored in the past and can be re-explored with the advent of modern in-roadway lighting, crosswalk lighting, vehicle headlamps, and education.

Vertical illuminance is among the most important factors for determining the contrast and visibility for a crossing pedestrian. Methods for achieving the required vertical illuminance may need to be explored because current standards are often believed to be difficult to achieve. The amount of vertical illuminance required may vary depending on an array of variables including environmental lighting, intersection lighting, crosswalk size, and crosswalk geometry.

In-roadway lighting is becoming an important area of interest as more agencies adopt it as a method of promoting safety and reducing energy. In-roadway lighting at crosswalks is believed to be a benefit for drivers to visualize crosswalks ahead of time and for pedestrians to have a lit path. However, installations of in-roadway are difficult to maintain and can be very expensive. As a result, the effects of in-roadway lighting in combination with area signage, overhead lighting, and retroreflective markings need be explored along with a benefit and cost assessment to better inform policy.

Newer technology for modern roadway lighting has led to more precise output from luminaires. This results in less light spilling over to sidewalks since higher-quality optics provide adequate lighting for the roadway but not for pedestrians. Investigating the impact of newer technologies, such as the Nadir Dump and other luminaire types, may inform future roadway lighting design with sidewalks and crosswalks in mind.

Seasonal Fluctuations in Daylight Hours and Effects of Daylight Savings Time

The change in when daylight occurs because of the changing seasons and the adjustments that occur with the use of Daylight Savings Time (DST) mean that there are periods when road users may be accustomed to daylight but experience darkness. In some locations and during certain times of the year, peak traffic volumes may occur at night. With DST, the sudden change in whether or not a commute occurs during daylight has significant safety implications, especially on pedestrians and cyclists.(40) These road users are often poorly visible due to an absence of lighting. If agencies provide street lighting, it may be valuable to investigate how they address DST changes and gradual seasonal daylight changes, whether with automatic lighting or with timers. Consistency in when lighting is provided relative to ambient light is important, especially for these vulnerable road users.

Glare from Beacons, Signals, and Other Lights

Flashing beacons and lights used on signs at intersections, midblock crosswalks, and other locations have been shown to be effective at grabbing attention, but glare from these lights may have a detrimental effect on drivers’ ability to see pedestrians and marked crosswalks, especially at night. Some text in the MUTCD instructs that automatic dimming for traffic signals should be used if a signal’s indication is bright enough to cause glare.(37) There appear to be no specifications, however, for appropriate levels of illuminance for these lights at night. In addition to flashing beacons used for crosswalks, the concern regarding glare may be relevant for all sources of light, including all traffic signals (including PHBs), LEDs on signs, railroad gates, in-roadway lights, and lane use signals.

Assumptions for Pedestrian Walking Speed

The assumed pedestrian walking speed suggested in the MUTCD is conservatively set at 3.5 fps, with accommodation for slower walking speeds for special conditions.(37) It is likely that the research used to select the recommended pedestrian walking speed focused exclusively on daytime conditions, neglecting the behavior of pedestrians at night. Researchers in a recent study showed that the time of day may impact walking speeds, with pedestrians walking approximately 0.5 fps slower during the evening peak than during the morning peak.(41) That analysis, however, involved data collected only during those two time periods. Additionally, the data were collected in New York City (Manhattan), where the density of pedestrians tends to be high. Regardless, a different walking speed attributed to the time of day may justify examining the issue with more depth.

School Zones Marked with Word Markings

There are several instances in the MUTCD that specify that certain objects must have retroreflective properties.(37) There is currently no language specifying that the “SCHOOL” word marking informing drivers they are entering a school zone must be retroreflective. Although it is likely that the marking would be made of the same retroreflective material as the nearby line markings, there may be a question about whether the marking should be held to a standard of maintained retroreflectivity. If there are periods of the year during which school activities begin or end during dark or nighttime conditions, there may be justification to evaluate the brightness of the marking.

Bike Lanes

FHWA has approved the use of green pavements to delineate bike lanes. Generally, there is no nighttime visibility component to the green bike lanes (adding glass beads reduces the friction). There is a need to develop technologies and materials that can help provide nighttime visibility to bike lanes so that their paths are as easily identified during dark conditions as they are during daytime conditions.

Other Areas with Nighttime Visibility Concerns

The following research concerns do not fit within the three target program areas, but the topics arose in the course of identifying gaps in current practices and research related to nighttime visibility. Several of these research areas are a result of budding and future technologies that promise ways to provide cheaper and more-efficient lighting through sensors and automation. The effect these technologies will have is unforeseen, and research into their impacts on safety is warranted.

Health Effects of Roadway Lighting

In a recent document, the American Medical Association (AMA) noted that the presence of the blue portion of the lighting spectrum in light sources has the potential to impact the health of humans living close to the light source.(42) AMA recommends that the light source be no greater than 3000K in color temperature. Studies have, however, shown that light sources with a 4100K color temperature perform at a higher level in terms of the visibility of objects in the roadway.(43) It is vital that these issues be considered.

Energy Consumption

Energy conservation has become a focus for many precincts and agencies. Removing lighting from infrastructure has become a popular method of conserving energy with some disregard to the impact of safety. Exploring methods of conserving energy in areas while maintaining safety should be a focus to inform policy and prevent uninformed decisions by infrastructure managers. Some known methods to efficient methods of lighting that allow for reduced energy consumption include the use of in-roadway lighting to highlight conflict points and prevent light trespass. Adaptive lighting technology is also a popular method of utilizing light only when it is necessary to do so.

The natural extension of this would be lighting on demand. LED lighting can be implemented with motion sensors or through connected-vehicle technologies to have lighting respond to the presence of a vehicle or pedestrian. Sample projects have been implemented in a controlled test road environment, but wider implementation will be developed in the future.

Automation

The anticipated adoption of connected and automated vehicles allows for innovative approaches to lighting. On one hand, vehicle automation may reduce the need to use lighting because so much of what the vehicles “see” will not be with light in the visible spectrum. On the other hand, there will still be other road users (such as pedestrians) that are just as deserving of safe and efficient transportation. The concept of smart cities with connected infrastructure opens up opportunities to apply lighting on demand and intelligently adapt to the users’ needs.

Lighting Uniformity

Uniformity is regarded as a desired effect of roadway lighting, though some research disputes this claim. There is room to explore the limits of lighting uniformity and its interaction with in‑roadway lighting, retroreflectivity, pedestrian visibility, and lighting color. The day-to-night effect of uniformity is important since daylight is highly uniform and roadway lighting generally strives to replace daylight; however, differences in source location and mount height tend to affect the angle of intensity of lighting. This causes severe shadows. In addition to roadway lighting, uniformity in intersections where light sources from other vehicles and infrastructure exist should be investigated.

Bridges

The AASHTO Green Book calls into question the justification of cost for lighting rural bridges. Whether or not rural bridges attribute to a number of conflicts and collisions that can be rectified by the addition or subtraction of lighting or delineators should be explored.(7)

Interaction of Overhead Lighting

The effect roadway lighting has on fauna, foliage, and crops is an aspect of lighting that is rarely documented in policy. Exploring the factors that benefit or harm wildlife in regard to lighting can provide insight on better and safer implementations. Research on this topic can also seek to understand the effect of environmental lighting, which is currently based on consensus knowledge. Adaptive lighting takes environmental effects into account, but the extent has not been properly researched.

Curve Design Accounting for Nighttime Line of Sight

One of the controls for the design of horizontal curvature is the line of sight cutting across the inside of the curve, limited by an obstruction that is offset from the road. The equation for calculating whether or not there is sufficient sight distance for curves is based on an assumption that the obstruction is the only limiting factor and the object would otherwise be visible to the driver. At night, however, driver vision can be restricted to the pattern of light from headlamps, which is mostly concentrated in a direct line in front of the vehicle. driver at night may not be able to see an object in the road near the end of a curve since the headlamps are only directed straight ahead. The limited amount of headlamp illumination that is scattered horizontally needs to be considered to properly judge sight distance when designing horizontal curves. Potential research in this area should evaluate the distribution of light from headlamps and whether objects would be visible at the wide viewing angles that can occur with sharp curves.

 < Previous Table of Content Next >
Page last modified on April 5, 2017
Safe Roads for a Safer Future - Investment in roadway safety saves lives
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000