TEMPORARY LOSSES OF HIGHWAY CAPACITY AND IMPACTS ON PERFORMANCE: PHASE 2

Title | Contents | Acknowledgements | Exec. Summary
1. Intro | 2. Approach | 3. Crashes | 4. Breakdowns | 5. Work Zones | 6. Weather | 7. Signal Timing
8. RR Crossings | 9. Toll Facilities | 10. PUD | 11. Results Summary | 12. Next Steps | 13. References

6.  Weather

Adverse weather conditions have a major impact on the operation of our Nation's roads.  Weather such as fog, rain, snow, and ice can reduce visibility and/or vehicle traction, causing drivers to reduce speeds and increase following distances. This reduces roadway capacity and increases delays. It can also cause drivers to cancel or re-schedule trips or re-route them around problem areas (e.g., roads prone to localized flooding or steep grades that become slick in icy weather), but these effects were outside the scope of this study. These events can also cause crashes or disable vehicles, further reducing capacity and increasing delays. However, it should be noted that the study did not attempt to determine crashes or disabled vehicles caused by these weather events nor did it attempt to consider the additive impacts of these events taking place during weather events—for example, it did not attempt to determine the additive delay caused from a crash occurring in a snowstorm.  Finally, rain, fog, snow, and ice are not the only weather events that can cause temporary delay. Solar glare, dust storms, gusting winds, flash floods, melting and re-freezing of snow and ice, and forest fires, as well as catastrophic events such as floods and hurricanes, can all reduce capacity and cause delay. However, due to data limitations and time and funding constraints, these events were not included in the study. Therefore, the weather-related delay estimates should be viewed in the context of these limitations in scope.

6.1  Methodology

Capacity reductions and delays from fog, snow, and ice were estimated in the initial phase of the TLC study. For TLC2, a revised methodology was used to estimate capacity reduction and delays for these weather phenomena, along with those resulting from rain. The primary difference in the methodologies was in the capacity reductions attributed to each type of weather event. In the initial TLC study, capacity reductions were based on estimates reported in a limited number of studies in the literature. However, for TLC2, newly available capacity and speed reduction estimates from FHWA's Road Weather Management Program were used.

The process for estimating delay for adverse weather consisted of the following tasks:

Step 1.  Identify all rain-, fog-, snow-, and ice-related weather events, along with the temporal and location characteristics of each event.  Location was defined in terms of public forecast zones or "region of influence" defined by weather stations. Archived weather databases compiled by the National Climatic Data Center (NCDC) were used to identify and locate events.

Step 2.  Map the weather events to the highway network system.  Some weather-related data follows county boundaries.  This is convenient since highway data from HPMS is available by county.  However, for weather data that is not county-based, a GIS-based computer model was used to map the weather data to the National Highway Performance Network (NHPN).

Step 3.  Estimate capacity losses on highways in impacted counties.  Capacity losses due to weather were based on weather-induced capacity and speed reductions estimated by FHWA's Road Weather Management Program.

Step 4.  Estimate the normal delay (congestion delay without adverse weather condition) for roadways in each county during the time of the event.

Step 5.  Estimate the delay with adverse weather conditions for roadways in each county during the time of the event.  This was accomplished by repeating the same delay estimation step with the appropriate capacity reduction for the weather type. For each link, if the average speed without the weather event was greater than the typical travel speed for that event, the delay was calculated based on the difference between those two speeds.  Otherwise, the delay was not adjusted.

Step 6.  Calculate the delay induced by adverse weather conditions for each county as the difference in the estimated delay with the adverse weather condition and the estimated delay without the adverse weather condition.

Step 7.  Calculate the delay induced by adverse weather conditions for each public forecast zone or "region of influence" defined by weather stations as the sum of the delay for all involved counties multiplied by the percentage of each county's major arterial mileage lying within the public forecast zone.

6.1.1  Identifying Adverse Weather Conditions

The first step in assessing the effect of adverse weather conditions on travel was to identify events with the potential to reduce capacities and cause delays.  Adverse weather events were identified using archived weather databases from the National Climatic Data Center (NCDC). The Center has long served the Nation as a national resource for climate information. 

NCDC's "Storm Data and Unusual Weather Phenomena" database was used to identify fog-and ice-related weather events. The Storm Data database contains a chronological listing, by state, of events such as hurricanes, tornadoes, thunderstorms, hail, floods, drought conditions, lightning, high winds, snow, temperature extremes, and other weather phenomena. Since the database identifies events in more-specific terms than used in the study, each fog-and ice-related event was assigned to more broad categories as shown in the table below.

In principle, the database only includes storms and other important climatological events having sufficient intensity to cause injuries, loss of life, and/or significant property damage.  "Significant" events, for the purposes of inclusion in the database, include those causing losses of at least $1,000.  Within the National Weather Service Weather Services Operation Manual (Chapter F-42, Storm Data and Related Reports), quantitative measures and procedures are provided to identify severe and/or significant weather and weather-related events.  However, much of the data is compiled by NWS personnel based on their judgment in ascertaining the severity of a meteorological event. Furthermore, it is possible for some events, especially fog, to have a significant impact on traffic without causing enough financial damage to be classified as severe.

In the database, the location and impacted area for an event is described in terms of (1) the county name and nearby town, or by mileage and direction from one town or between two towns, or (2) public forecast zones.  Public forecast zones are geographical areas within a state designated by the Weather Service Forecast Offices (WSFO) and Next Generation Radar (NEXRAD) Weather Service Forecast Offices (NWSFO) with consent of the Regional Headquarters.  Ideally, a forecast zone is an area with sufficient climatological and meteorological homogeneity to allow a single forecast to serve as the local forecast for the communities within the area.  Each WSFO's area of responsibility is divided into forecast zones.  The only exceptions are a few mountainous or largely unpopulated areas of the far western United States, certain wilderness areas in Alaska, island areas of the Pacific Region, and Isle Royale in Lake Superior.

Zone boundaries are typically determined by considering:

• Homogeneity of climate and usual weather features
• Population distribution
• Local communications and dissemination capabilities

In practice, most public forecast zones follow county boundaries, although some counties might be split into multiple public forecast zones and some forecast zones might consist of multiple counties.  For example, public forecast zone boundaries in Alabama typically coincide with county boundaries.  This is true for all counties except Baldwin and Mobile counties, which border the Gulf of Mexico.  The local WSFO further divides these two counties into four public forecast zones.  However, in some states, public forecast zone boundaries do not follow county boundaries.  For example, public forecast zones in Utah and Colorado typically span several counties.  Each public forecast zone is assigned a Universal Generic Code. This allows local offices to create their own "regional" areas.

Rain and Snow

Rain-and snow-related weather events were identified using the National Weather Service's "Hourly Precipitation Data (HPD) TD3240" and "Cooperative Summary of Day TD3200" databases. The HPD database contains hourly precipitation amounts recorded by rain gages located at National Weather Service, Federal Aviation Administration, and cooperative observer stations. HPD includes maximum precipitation for nine (9) daily periods, ranging in length from 15 minutes to 24 hours, for selected stations.

Unfortunately, the HPD only contains precipitation information; it has little information on the type of precipitation, such as rain, snow, or freezing rain. Therefore, the Summary of Day database was used to infer the precipitation type. The Summary of Day database includes daily characteristics such as maximum/minimum temperatures, precipitation, and snowfall/snow depth. Some stations have additional data such as evaporation, soil temperature, peak wind gust, etc. This data is a compilation of daily observations initially obtained from state universities, state cooperatives, and the National Weather Service. Most stations that collect HPD data also collect Summary of Day data. However, for the small percentage that did not collect this data during the study year, data from the nearest station that did collect it was used.

A few assumptions were made to infer precipitation type. If the Summary of Day data gives an amount in the "snow/ice pellets" field, it was assumed that all of the precipitation for that day was snow. The precipitation amount in the HPD was also assumed to be snow if the maximum temperature as given in the Summary of Day data was 32ºF or less. Another limitation of the data is the accuracy of precipitation gauges. These gauges measure snow and ice somewhat less accurately than rain.

The HPD and Summary of Day databases contain information on the locations of the cooperative weather stations (i.e., the point locations where the weather measurements are made). However, the weather data for these one-dimensional points must be applied to two-dimensional areas. This was accomplished by creating Thiessen Polygons around the weather station location points. Thiessen polygons, also referred to as the Dirichlet Tessellation or the Voronoi Diagram, define the individual "region of influence" around each point within a set of points. Thiessen polygons are polygons whose boundaries define the area that is closest to each point relative to all other points. Thiessen polygons, which are generated from a set of points, are mathematically defined by the perpendicular bisectors of the lines between all points.

By using the Thiessen polygons, this study assumes the adverse weather condition is represented by the weather data collected at the cooperative weather station at the centroid and that the weather is the same throughout the polygon. This assumption is adequate if a sufficient number of cooperative weather stations are available and the polygons are small enough to describe weather phenomenon effectively. Some of the resulting polygons in the TLC2 study are somewhat large due to the sparsity of weather stations in some areas. However, these are typically lower population areas that are likely to generate less traffic. Therefore, any inaccuracies in weather impacts within these larger polygons should not greatly affect delay estimates. The figure below shows the Thiessen polygons used with this data set.

Fig. 14. Weather areas defined in terms of Thiessen polygons

6.1.2  Mapping Adverse Weather Conditions to Highway Segments

FHWA's HPMS was used to help estimate weather impacts in terms of traffic capacity losses and delays. HPMS provides traffic operation information by highway functional class, county, and state. Since the locations of fog- and ice-related weather events are described using public forecast zones and rain- and snow-related events are assigned to Thiessen polygons, additional effort is needed to map these weather conditions to HPMS data, which is county-based.  To calculate weather impacts, the percentage of each county's major arterial mileage affected within a given public forecast zone and/or Thiessen polygon was determined for each public forecast zone and Thiessen polygon.

Because the HPMS does not include coordinate data (i.e., longitude and latitude), an additional database was used to help map the coordinate-based weather data to highway segments.  FHWA's National Highway Planning Network (NHPN) Ver. 2.2 was used to establish linkage between weather data and counties (U.S. DOT/FHWA 2000).  The NHPN, which has been under development since the mid 1980s, was originally assimilated from a variety of sources at a nominal scale of 1:2 million and contains a set of data attributes that are suited to analytical modeling of large-scale transportation activities.  The accuracy of the Version 2 database has changed from a scale of 1:2 million to a scale of 1:100,000 (an accuracy of about 80 meters rather than 1,500).

The method for mapping highway links to coordinate-based weather data consisted of the following steps:

Step 1. "Cut" All NHPN Links at Public Forecast Zone and Thiessen Polygon Boundaries.

All NHPN links (roadway segments) are already "cut" at county boundaries.  In order words, each NHPN link lies entirely within a county boundary.  In order to establish linkage with coordinate-based weather data, all NHPN major arterial links were "cut" at public forecast zone and Thiessen polygon boundaries.

Step 2. Aggregate All NHPN Major Arterial Links by Public Forecast Zone and Thiessen Polygon.

Aggregated mileages can be calculated for each public forecast zone and each Thiessen polygon based on the NHPN major arterial link's public forecast zone and Thiessen polygon subdivision.

Step 3. Determine Each County's Percentage of Major Arterial Mileage Lying within Each Forecast Zone and Thiessen Polygon.

Using the information obtained from the previous step, the percentage of each county's major arterial mileage within each public forecast zone and Thiessen polygon was determined.

6.1.3  Estimating Capacity Losses

Adverse weather conditions affect capacity and reduce operating speeds significantly, and each type of weather affects highway travel differently.  The Transportation Research Board's Highway Capacity Manual discusses the traffic operations impacts of rain, snow, and fog and provides quantitative capacity and speed reduction factors based on limited studies—quantitative information on ice is not provided. For the initial study, information from the Highway Capacity Manual and studies by Lamm et al. (1990), Ibrahim and Hall (1994), Hogema et al. (1994), Aron et al. (1994), and Brilon and Ponzlet (1995) were used to produce estimates of reduced capacity. The revised study used capacity reduction factors compiled from various sources and studies by the Federal Highway Administration's (FHWA's) Road Weather Management Program.

The Road Weather Management Program was established by the FHWA Office of Operations to facilitate deployment of integrated road weather systems, decision support applications, and tools and practices in response to adverse weather that meet the needs of all transportation system users. Under this program, the Office of Operations has collected and compiled, from several sources, quantitative weather impacts on different roadway facilities at different locations. These impacts are summarized in the following table.

The estimates in the above table do not cover all weather types used in this study. Therefore, it was necessary to make some assumptions. For example, it was assumed that fog impacts would be the same as for situations where visibility on rural freeways was less than 530 feet. These values were used for all highway types. Furthermore, it was assumed that the effects of ice were the same as those for heavy snow.

The above table provides mobility reduction estimates for heavy and light precipitation, although it does not quantitatively define the differences between "heavy" and "light." This study assumes rain or snow falling at a rate of 1 inch or more per hour to be heavy, and assumes other amounts are light. The highway capacity and speed reduction impacts of adverse weather conditions used in the study are summarized below.

To accurately estimate capacity reduction, the time period during which each event impaired visibility and affected roadway conditions must also be determined. Unfortunately, there is no available national-level data on the length of time weather events impacted roadways. Therefore, the duration of the weather event was used as a surrogate for the duration of the capacity loss. This assumption is valid for events that only affect visibility, such as fog. However, it correlates less closely with capacity reductions realized from adverse pavement conditions, since events such as rain, snowfall, and ice can impede traffic well after precipitation stops. Furthermore, if the road temperature is sufficient, snow may not stick on the pavement and may cause little capacity reduction. Therefore, this assumption likely leads to an underestimate of capacity reduction and delay.

6.1.4  Estimating Delay

All methodologies used to estimate delays are based on procedures outlined in Highway Capacity Manual 2000 (TRB 2000).  These are described in the sections below.

Rural/Urban Interstates and Urban Other Expressways

One of the major traffic operation parameters required for calculating traffic delay is average speed.  The average speed information can be determined by the speed-volume relationship.  The traffic volume by hour of the day and day of the week can be estimated based on the AADT and the k-factor (peak-hour traffic volume factor) from HPMS.

Traffic flow on a freeway segment can be categorized into three flow types:  under-saturated flow, queue discharge flow, and over-saturated flow (Fig. 15). Each flow type is defined within general speed-flow-density ranges, and each represents different conditions on the freeway.

• Under-saturated flow represents traffic flow which is unaffected by upstream or downstream conditions.  This flow regime is generally defined within a speed range of 90 to 120 kilometers per hour (kph) at low to moderate flows and between 70 and 100 kph at high flow rates.
• Queue discharge flow represents traffic flow that has just passed through a bottleneck and is accelerating back up to the free-flow speed of the freeway.  Queue discharge flow is characterized by relatively stable flow as long as the effects of another bottleneck downstream are not present. This flow type is generally defined within a narrow range of flows, 2,000–2,300 pcphpl, with speeds typically ranging from 55 kph up to the free-flow speed of the freeway segment.  Lower speeds are typically observed just downstream of the bottleneck.  Depending upon horizontal and vertical alignments, queue discharge flow usually accelerates back up to the free-flow speed of the facility within 1 to 2 kilometers downstream from the bottleneck.  Studies suggest that the queue discharge flow rate from the bottleneck is lower than the maximum flows observed prior to breakdown.  A typical value for this drop in flow rate is approximately 5 percent.
• Over-saturated flow represents traffic flow that is influenced by the effects of a downstream bottleneck.  Traffic flow in the congested regime can vary over a broad range of flows and speeds depending upon the severity of the bottleneck.  Queues may extend several kilometers upstream from the bottleneck.  Freeway queues differ from queues at intersections in that they are not static, or standing.  On freeways, vehicles move slowly through a queue, with periods of both stopping and movement.

Speed-flow and density-flow relationships for a typical basic freeway segment under either base or non-base conditions in which free-flow speed is known are shown in Fig. 16. Recent freeway studies indicate that speed on freeways is insensitive to flow in the low to moderate range. This is reflected in Fig. 16, which shows speed to be constant for flows up to 1,300 pcphpl for a 120-kph free-flow speed. For lower free-flow speeds, the region over which speed is insensitive to flow extends to even higher flow rates.

Under base traffic and geometric conditions, freeways will operate with capacities as high as 2,400 pcphpl. This capacity is typically achieved on freeways with free-flow speeds of 120 kph or greater. As the free-flow speed decreases, there is a slight decrease in capacity.  For example, capacity of a basic freeway segment with a free-flow speed of 90 kph is expected to be approximately 2,250 pcphpl.

Figure 15.  Freeway traffic flow types

Fig. 16.  Speed-flow relationships for a typical basic freeway segment

As indicated in Fig. 16, the point at which an increase in flow rate begins to impact the average passenger car speed varies from 1,300 to 1,750 pcphpl. Speed will be reduced beginning at 1,300 pcphpl for freeway segments with a free-flow speed of 120 kph.  For facilities with lower free-flow speeds, the average speed begins to diminish at higher flow rates.

The relationships in Fig. 16 were digitized and transformed into speed volume-to-capacity ratio relationships.  Polynomial equations to the fourth order were used and "fitted" to these curves (Equations 14-17 and Fig. 17). These equations made it possible to use a computer program to calculate average speeds based on volume-to-capacity ratio.

Using the speed limit (used as free flow speed), AADT, peak-hour factor, directional flow split (applied to morning and afternoon peak-hours) and peak-hour capacity information from HPMS, the average speed was estimated based on these equations.  If the volume-to-capacity ratio was larger than 1, the queue was tracked, and queue length and queue delay were calculated.  The total travel time was calculated as the travel time (based on average speed estimated by one of the abovementioned equations) plus the queue delay.

Figure 17.  Speed relationship to volume/capacity ratio for basic freeway segment

Equations 14-17:

Avg Speed (mph)_FFS=75mph = -46.97(v/c)⁴ + 22.72(v/c)³ + 5.0909(v/c)² - 2.1844(v/c) + 74.6

Avg Speed (mph)_FFS=68mph = -87.69(v/c)⁴ + 107.92(v/c)³ - 40.719(v/c)² + 4.5651(v/c) + 68.345

Avg Speed (mph)_FFS=60mph = -111.01(v/c)⁴ + 164.11(v/c)³ - 74.212(v/c)² - 9.9859(v/c) + 60.091

Avg Speed (mph)_FFS=55mph = -79.645(v/c)⁴ + 717.36(v/c)³ - 62.256(v/c)² - 9.0488(v/c) + 55.87

Rural/Urban Other Major Principal Arterials

Intersections without Signals or Stop Signs:  The speed-flow relationships for a typical uninterrupted-flow segment on a multilane highway under either base or non-base conditions in which free-flow speed is known are shown in Fig. 18.  The operating characteristics for a multilane highway may be slightly lower than for a freeway because drivers on multilane highways allow for potential conflicts with turning traffic, even when there are no access points in the immediate vicinity.

As indicated in Fig. 18, the speed of traffic on a multilane highway is insensitive to traffic volume up to a flow rate of 1,400 pcphpl.  The exhibit shows that the capacity of a multilane highway under base conditions is 2,200 pcphpl for highways with a 100-kph free-flow speed.  For flow rates from 1,400 to 2,200 pcphpl, the speed on a multilane highway with a 100-kph free-flow speed drops 12 kph.

Figure 18.  Speed-flow relationships on multi-lane freeways

The capacity value of 2,200 pcphpl is representative of the maximum 15-minute flow rate that can be accommodated under base conditions for highways with a free-flow speed of 100 kph. Actual capacities on specific multilane highway sections may vary from this value.

The relationships presented in Fig. 18 were digitized, polynomial equations to the fourth order were "fitted" to the displayed curves (Equations 18-21 and Fig. 19).  These equations made it possible to use a computer program to calculate average speeds based on volume-to-capacity ratio.

Equations 18-20:

Avg Speed (mph)_FFS=62mph = -2.2238(v/c)⁴ - 23.011(v/c)³ + 21.701(v/c)² - 4.3551(v/c) + 62.184

Avg Speed (mph)_FFS=55mph = -6.4985(v/c)⁴ - 10.795(v/c)³ + 14.347(v/c)² - 3.226(v/c) + 55.962

Avg Speed (mph)_FFS=50mph = 0.246(v/c)⁴ - 14.159(v/c)³ + 12.523(v/c)² - 2.4503(v/c) + 49.729

Avg Speed (mph)_FFS=44mph = -13.636(v/c)⁴ + 20.225(v/c)³ - 9.1272(v/c)² + 1.2136(v/c) + 43.496

Similar to the method used for freeways, the average speed was estimated using these equations and the speed limit (used as free flow speed), AADT, peak-hour factor, directional flow split (applied to morning and afternoon peak-hours) and peak-hour capacity information from HPMS.  If the volume-to-capacity ratio was larger than 1, the queue was tracked, and queue length and queue delay were calculated. The total travel time was calculated as the travel time (based on average speed estimated by one of the abovementioned equations) plus the queue delay.

Figure 19.  Speed relationship to volume/capacity ratio for multi-lane freeways

Signal-Controlled Intersections:  For principal arterials with signal-controlled intersections, the capacity for the arterial segment was calculated as the capacity at the signal-controlled intersections.  Thus, average speed was dominated by the delay time at these signal-controlled intersections.

The methodology described in Section 7.1.2, "Estimating Total Delay for Signal-Controlled Intersections," was used to estimate total delay time at signal-controlled intersections during adverse weather conditions. If the volume-to-capacity ratio was larger than 1, queue was tracked and queue length and queue delay were calculated.  The total travel time was calculated as the sum of the travel time (based on free flow speed), signal-controlled intersection delay time, and queue delay.

Stop-Sign-Controlled Intersections:  For principal arterials with stop-sign-controlled intersections, the capacity for the arterial segment was calculated as the capacity of the stop-sign-controlled intersections. Thus, average speed was dominated by the delay time at these intersections.

The delay experienced by a motorist is made up of a number of factors that relate to control, geometry, traffic, and incidents.  Total delay is the difference between the travel time actually experienced and the reference travel time that would result during ideal conditions, in the absence of incident, control, traffic, or geometric delay. This study quantifies only that portion of total delay attributed to traffic control measures (i.e., either traffic signals or stop signs).  This delay is called control delay.  Control delay includes initial deceleration delay, queue move-up time, stopped delay, and final acceleration delay.  With respect to field measurements, control delay is defined as the total elapsed time from when a vehicle stops at the end of the queue until the vehicle departs from the stop line.  This total elapsed time includes the time required for the vehicle to travel from the last-in-queue position to the first-in-queue position, including deceleration of vehicles from free-flow speed to the speed of vehicles in queue.

Average control delay for any particular minor movement is a function of the capacity of the approach and the degree of saturation.  The analytical model used to estimate control delay (Equation 22) assumes that the demand is less than capacity for the period of analysis.  In situations where the degree of saturation is greater than about 0.9, average control delay is significantly affected by the length of the analysis period. In most cases, the recommended analysis period is 15 minutes.  If demand exceeds capacity during a 15-minute period, the delay results calculated by the procedure may not be accurate.  In this case, the period of analysis should be lengthened to include the period of over-saturation.

The constant value of 5 seconds/vehicle is included in Equation 22 to account for the deceleration of vehicles from free-flow speed to the speed of vehicles in queue and the acceleration of vehicles from the stop line to free-flow speed.

Equation 22:

where

d = stop sight control delay in seconds/vehicle

v_x = flow rate for movement x in vehicles/hour

c_x = apacity of movement x in vehicles/hour

T = analysis time period in hours (T = 0.25 for a 15-minute period)

If the volume-to-capacity ratio was larger than 1, the queue was tracked, and queue length and queue delay were calculated.  The total travel time was calculated as the travel time (based on free flow speed, stop-sign-controlled intersection delay time) plus the queue delay.

To estimate the delay due to adverse weather conditions, the normal delay (congestion delay without the adverse weather condition) for roadways in each county during the time of the event was estimated.[7]  The delay with adverse weather conditions for roadways in each county during the time of the event was then estimated.  This was accomplished by repeating the same delay estimation step with the appropriate capacity reduction for the event type.  If the link average speed was greater than the typical travel speed for that event, the delay was calculated at that speed.  Otherwise, the delay was not adjusted.  The delay induced by adverse weather conditions for each county was calculated as the difference in the estimated delay with the adverse weather condition and the estimated delay without the adverse weather condition. The delay induced by adverse weather conditions was then calculated for each weather zone (public forecast zone or Thiessen polygon) as the sum of the delay for all involved counties multiplied by the percentage of each county's major arterial mileage lying within that zone.  These estimates were summed to produce a national total.

6.2  Results

The methodology and data used in TLC2 to estimate weather-related capacity reductions and delay was modified from the original TLC study. Therefore, this section includes the revised weather-related delay estimates, along with an illustration of the difference between the estimates produced by both methods and an explanation as to why some of the revised estimates are considerably different from those in the original study.

6.2.1 TLC2 Results

The TLC2 study estimated that, in 1999, rain, fog, snow, and icy conditions temporarily reduced capacity on freeways and principal arterials by approximately 20.9 billion vehicles.  This resulted in an estimated 330.1 million vehicle-hours of delay. Rain accounted for most of the delay from adverse weather (71 percent), followed by ice (14 percent), snow (13 percent), and fog (2 percent). Urban areas experienced 92 percent of the delay from weather events, with most of this delay experienced on principal arterials. As shown in the figures below, capacity reductions were less likely to translate into delays in rural areas.

Figure 20.  Most weather-related capacity reduction occurred from rain on urban and rural arterials.

Figure 21.  Most weather-related delay was experienced in urban areas

6.2.2 Comparing TLC and TLC2 Results

The TLC2 study's estimated capacity reduction and delay resulting from snow and fog are significantly different from those estimated by the initial TLC study. This is primarily due to two factors. First, different speed reduction factors were used in these studies. In TLC2, speed reduction factors found on the FHWA Road Weather Management Program website were used in lieu of those derived from the Highway Capacity Manual. Second, the Road Weather Management Program's site provided separate factors for "heavy" and "light" weather conditions, whereas the sources used for the initial TLC only applied to "heavy" conditions. In the first TLC study, all events were assumed to be heavy. Conversely, a significant number of the weather events in the TLC2 study were characterized as light fog or light snow. Therefore, these lower speed reduction estimates affect a large number of the weather events within the study, resulting in much lower delay estimates for these kinds of events and for the overall delay from weather. These are the only differences in the methodologies for estimating impacts from fog, snow, and ice.

Fig. 22. Capacity reductions due to fog and snow estimated by TLC2 were significantly lower than those estimated by the initial TLC study.

Fig. 23. Delays due to fog and snow estimated by TLC2 were significantly lower than those estimated by the initial TLC study.

6.3 Reliability

6.3.1 Methodology

Established analytical procedures were used to map weather information to HPMS data. However, due to the complex nature of local weather patterns and the factors that determine them, these methods are qualified as having a medium degree of confidence.

Methodologies from the Highway Capacity Manual were used to estimate capacity loss due to speed reductions and delay due to capacity reductions. These methodologies have evolved over the years and have been updated and enhanced continuously when new information or methodologies are made available.  They are well established and accepted within the traffic engineering community. These methodologies are qualified as having a high degree of confidence.

6.3.2 Data & Key Assumptions

The "Hourly Precipitation Data (HPD) TD3240" and "Cooperative Summary of Day TD3200" databases compiled by National Weather Service are qualified as having high degree of confidence. However, it should be noted that precipitation gauges are not as accurate for frozen precipitation and, therefore, may slightly underestimate snow measurements.

The "Storm Data and Unusual Weather Phenomena" database compiled by the National Climatic Data Center (NCDC) is qualified as having a high degree of confidence. However, as noted previously, the database only includes storms and other important climatologic events having sufficient intensity to cause injuries, loss of life, and/or significant property damage ($1,000 or more). Less severe events can still reduce capacity and cause delay―this is especially true for fog, which can cause delay even if no significant monetary damages result. However, due to data limitations, these events are not modeled and their overall contribution to delay is not known.

Adverse weather impacts on capacity and operating speeds are based on information acquired from the FHWA Office of Operations, Road Weather Management Program web site. However, this information was not collected by FHWA, and FHWA has a low degree of confidence in its reliability.

Several assumptions were made in modeling weather and its impacts that have a significant influence on the reliability of these capacity loss and delay estimates.

• The TLC2 study assumes that weather-related impacts occur only during the time that the event occurs (for rain, ice, and show, this would be the duration that precipitation fell). While this relationship rarely exists, especially for ice-and snow-related events that can affect capacity for hours or days after precipitation falls, there is no national-level data on how long roadways were affected by weather events. In addition, the duration of impacts is quite localized and depends upon factors other than the weather itself, such as the availability and use of equipment to prevent or mitigate snow and ice impacts. Therefore, in the absence of data, we have decided to limit the duration of impact to the duration of precipitation, the only part of the weather event for which data was available. Therefore, this assumption is accorded a low degree of confidence for ice and snow.
• The precipitation data does not include specific information about the precipitation form (rain, snow, etc.). Therefore, the precipitation type (rain vs. snow) was estimated based on temperature: at a temperature of 32ºF or less, precipitation was assumed to be in the form of snow. This assumption is accorded a medium degree of confidence.
• Capacity reduction estimates for snow were used as a surrogate for ice-related events since no estimates were available for ice. It was assumed that the impacts of these events were at least similar in nature. While ice likely reduces capacity more than snow, snow-related capacity reductions were used in keeping with the study's philosophy of erring on the conservative side.
• The Road Weather Management Program data provides speed reduction and capacity reduction data for "heavy" and "light" weather conditions, but the definitions of heavy and light are not quantified. Therefore, any precipitation falling at a rate of over 1 inch per hour was classified as heavy. This assumption is accorded a medium degree of confidence.

The same traffic demand data and surface street assumptions used to estimate crash delays were used for breakdowns. Therefore, the same caveats apply: traffic demand data is accorded a low level of confidence and surface street characteristics are accorded a medium level of confidence (see section 3.3.2).

As indicated at the beginning of this report, TLC2 does not consider the impact of rescheduled or canceled trips, although they have a significant impact on demand in some cases. Also, a number of weather events that can cause delay were not included in the report due to data limitations and/or limitations in time and funding. These include solar glare, high winds, dust storms, flash flooding, and melting and re-freezing of ice and snow, along with catastrophic events such as hurricanes and floods. Finally, the increased likelihood of crashes and disabled vehicles is not modeled nor are the additive impacts of these events occurring during a weather event.

The weather-related estimates should be viewed in the context of these limitations.

7. It should be noted that the event duration was assumed to be the time period during which the weather event took place. For rain, snow, and ice, the duration would include only the time during which precipitation fell; it would not include the impact of the precipitation on the ground after it stopped falling.

Previous | Table of Contents | Next