Chapter 4. Traffic Flow Parameters in Varying Weather Conditions
Data Collection
Some traffic flow characteristics change in inclement weather. In this study, observations are made of measured saturation flow, vehicle speed, and start-up lost in a range of weather conditions. These measurements are used first to determine how to change each parameter when developing the special timing plan and second, to determine under what conditions a special timing plan should be in operation. The measured parameters are defined as follows (FHWA 1994):
- Saturation Flow Rate: The maximum rate of traffic flow for a single lane under prevailing conditions if it has 100 percent effective green time (FHWA 1994)
- Lost Time: The time that an intersection is not being used by any movement. This occurs at the beginning of a movement and during the clearance intervals (FHWA 1994).
- Free-flow speed: The uninhibited speed of a vehicle on a length of roadway.
A range of seven weather severity categories is defined in Table 4.1. These are used consistently throughout the data collection and analysis and correspond to the categories used in the FHWA 1977 study.
Scale of Road-Surface Weather Conditions
| Severity ID | Description |
| 1 | Normal/Clear |
| 2 | Rain |
| 3 | Wet and Snowing |
| 4 | Wet and Slushy |
| 5 | Slushy in Wheel paths |
| 6 | Snowy and Sticking |
| 7 | Snow Packed surface |
Saturation flow and speed data was collected during all available weather events over the winter season of 1999-2000. For comparison purposes, data also was collected in dry weather. On 14 different inclement weather days, more than 30 hours of saturation flow and free-flow speed information was collected. Data collection was collected primarily during the morning or evening peak hours. Due to unusually mild conditions during this season, there were only a few heavy snowstorms that occurred during peak hours, which limited data collection.
The two intersections selected for all data collection are: 700 East / 900 South and 1300 East / 500 South (Figure 4.1). The intersection geometries are shown in Figures 4.2-4.3. These two intersections were selected because they are on major corridors and near the University of Utah campus: 700E and 1300E are in the North-South direction with peak directional traffic into downtown (North) during the a.m. peak and from downtown (south) during the p.m. peak. Saturation flows were collected using JAMAR® TDC-8 traffic data collectors. Free-flow speeds were recorded with radar guns, while measuring the saturation flow.
Figure 4.1 Study Area
Figure 4.2 Intersection 1 (700 E & 900 So)
Figure 4.3 Intersection 2 (1300 E & 500 So)
Saturation Flow
Saturation flow decreases during inclement weather because of larger headways, slower speeds, and decreased acceleration rates. The saturation flow decreased as storm severity increases. The average measured values are shown along with the percent reduction from the dry saturation flow as an average of all four approaches in Table 4.2. The percent deviation is shown graphically on Figure 4.4. The raw saturation flow values and the average values by approach are shown in Appendix B.
Saturation Flow
| Road Surface Condition | Severity | 700 E 900 S | 1300 E 500 S | Average % Reduction | ||
| (AM) | (PM) | (AM) | (PM) | |||
| Dry | 1 | 1881(100%) | 1736 (100%) | 1752 (100%) | 1902 (100%) | 0% |
| Rain | 2 | 1680(89%) | 1711 (99%) | - | - | 6% |
| Wet and Snowing | 3 | 1751 (93%) | 1708 (98%) | 1491 (85%) | 1691 (89%) | 11% |
| Wet and Slushy | 4 | -a | 1476 (85%) | 1321 (75%) | 1647 (87%) | 18% |
| Slushy in Wheel Paths | 5 | - | 1421 (82%) | - | - | 18% |
| Snowy and Sticking | 6 | - | - | 1395 (80%) | - | 20% |
| Snow Packed | 7 | - | - | - | - | - |
aNo data available
Figure 4.4 Average Saturation Flow Reductions by Weather Condition
No storms during this winter season were severe enough to be classified as severity category seven. Sufficient saturation flow data was collected for all other categories. It is important to note that the largest drop in percentage between two adjacent categories (3 and 4) is 9 percent. The maximum reduction in saturation flow was 20 percent, which occurred during severity category 6.
It is useful to compare our results to those found in similar studies. Three other studies also include saturation flow measurements in inclement weather. These are: Fairbanks, Alaska; Minneapolis, Minnesota; and Anchorage, Alaska. Each of these studies uses a slightly different definition for inclement weather conditions. The Anchorage study defines summer as the time when temperatures are above 14º F, on dry roads, or above 32º F on wet roads (without ice). Winter is defined as the time of year when temperatures range from -22º F to 14º F with dry pavement or with well-sanded hard-packed snow on the road. Extreme means the air temperature is below -22º F or during snowfall, blizzard, and/or freezing rain, resulting in slippery roads and reduced visibility. The Minneapolis study defines inclement as a storm with an accumulation of three inches or more. The Fairbanks study defines inclement as when residual ice, snow, or frost on the road surface due to a storm slows traffic. although the definitions of inclement are slightly different for each study, we will show a comparison to observe that our results are in a similar range as other studies. The measured SLC inclement value of 1,432 is the mean saturation flow found between conditions 4-6. From saturation flow, a calculated headway change can be determined.
Comparison of Saturation Flow Reduction
| Weather Condition | Salt Lake City | Flow Rates at signalized intersections in cold weather (Botha and Kruse 1992) Fairbanks, AK | Anchorage Signal timing report (Bernardin, Lochmuller and Associates1995) Anchorage, AK | Adverse Weather Traffic Signal Timing (Maki 1999) Minneapolis, MN |
| Normal/clear | *1,808 | 1,792 | 1,816 | 1,800 |
| Inclement | 1,432 | 1,538 | 1,600 | 1,600 |
| Reduction | 21% | 14% | 12% | 11% |
* values in vehicles per second
Comparison of Headway Increase
| Weather Condition | Salt Lake City | Flow Rates at signalized intersections in cold weather (Botha and Kruse 1992) Fairbanks, AK | Anchorage Signal timing report (Bernardin, Lochmuller and Associates 1995) Anchorage, AK | Adverse Weather Traffic Signal Timing (Maki 1999) Minneapolis, MN |
| Normal/clear | *1.99 | 2.00 | 1.98 | 2.00 |
| Inclement | 2.51 | 2.34 | 2.25 | 2.25 |
| Increase | 26% | 17% | 14% | 13% |
* values in seconds per vehicle
Speed
Free-flow speeds decrease during rain or snowstorm events. Speeds were collected during dry and inclement weather. Below are the average values of free-flow speeds collected on 18 different days during the winter of 1999-2000. Speed data was usually collected during the peak (AM or PM) hours at each intersection. There is no data, however, for speeds beyond condition 5 for either intersection. The complete set of speed data for each specific intersection approach and time is listed in Appendix C. The average speeds for each intersection are provided in Table 4.5. Figure 4.5 shows the percent reduction in speed from the dry condition in graphical form.
Average Speeds and Percent Reduction from Dry Condition
| Road Surface Condition | Severity | 700 E 900 S | 1300 E 500 S | Average % Reduction | ||
| (AM) | (PM) | (AM) | (PM) | |||
| Dry | 1 | 39.0(100%) | 31.4(100%) | 28.4(100%) | 27.4(100%) | 0% |
| Rain | 2 | 34.3( 88%) | - | - | 25.2(92%) | 10% |
| Wet and Snowing | 3 | 31.4(81%) | 29.4( 93%) | - | 23.5(86%) | 13% |
| Wet and Slushy | 4 | -* | 22.0(70%) | - | 21.8(79%) | 25% |
| Slushy in Wheel Paths | 5 | 25.5(65%) | 23.4(75%) | - | - | 30% |
| Snowy and Sticking | 6 | - | - | - | - | - |
| Snow Packed Surface | 7 | - | - | - | - | - |
*No data available
Figure 4.5 Average Speed Reductions by Condition
It is important to note that the largest drop in speeds is between conditions 3 and 4 (about 17 percent). This is consistent with the largest drop in saturation flows, which also saw the largest decrease between conditions 3 and 4. By condition 5, there was an average decrease in speeds of about 30 percent.
This decrease is consistent with two other studies. (Maki 1999) found a decrease in speeds of about 40 percent during inclement weather. (FHWA 1977) found that interstate speeds are reduced by 36 percent during inclement weather.
Table 4.6 shows the collected data of this study compared to the FHWA (1977) findings.
Speed Data Comparison to FHWA (1977)
| Condition | Severity | Percent Reduction (This Study) | Percent Reduction FHWA (1977) |
| Dry | 1 | 0% | 0% |
| Wet | 2 | 10% | 0% |
| Wet and Snowing | 3 | 13% | 13% |
| Wet and Slushy | 4 | 25% | 22% |
| Slushy in Wheel Paths | 5 | 30% | 30% |
| Snowy and Sticking | 6 | - | 35% |
| Snowing and Packed | 7 | - | 42% |
The similarity in the results of the FHWA (1977) study indicates that our results are consistent with previous findings and increases our confidence in the data.
Start-Up Lost Time
Start-up lost time is based on the first four to five vehicles in a stopped queue. By the sixth vehicle, headways become more constant. The assumption is that lost time will increase with inclement weather. The start-up speeds of each queue (procession of cars) will be much slower during inclement weather because vehicles will have less tire traction, thus stalling their initial movements. The Anchorage, 1995 study found that inclement start-up times were equivalent to the summer conditions. They estimated that the additional time is accounted for in the saturation flow reductions and therefore no change to lost time was included in the revised signal timing. In Fairbanks, there was a small reduction but not appreciable from the two seconds recommended by the Highway Capacity Manual, 1985. MinnDOT (1999) found a 50 percent increase in start-up lost time from two to three seconds.
Based on our observations, the start-up lost time increased by an average of 23 percent from 2.0 to 2.46 seconds that corresponds to the increased headway for inclement saturation flow rates. This is based on 112 dry weather samples and 134 snowy (condition 4-6) samples. The rain (condition 2) seemed to have little impact on start-up lost time, as the 35 samples indicated that the average lost time changed from 2.0 seconds (dry) to 2.1 seconds (wet in conditions 2 and 3).
In addition to the start-up lost time, consideration must be given to the dilemma zone and amber time, all-red time and pedestrian crossing time.
Pedestrian Crossing Time
Knoblaugh et al (1996) identified that pedestrians, both young (under 65 years old) and older (65 years or older), increased their walking speed during inclement weather. Younger pedestrians increased by 9 percent from 4.82 feet per second (fps) to 5.24 fps while older pedestrians increased by 8 percent from 4.03 to 4.37 fps. Based on this research, the minimum crossing time to pedestrians should not change and therefore no impact to minimum signal timing restrictions exists.
Dilemma Zone, Amber and Red Time
While the discussion of start-up lost time indicates that reductions in intersection efficiency result, other factors also cause a decrease in intersection performance. Gap acceptance is reduced for permitted left turn movements primarily because of reduced traction. The updated 1997 HCM identifies the critical gap for left turns as being 4.1 seconds. While a gap acceptance study was outside the scope of this study, it clearly increased for severity levels 4 through 6. It is our opinion that there is a probable increase of 25 to 30 percent in critical gap time. There is, however, no substantial data to support this opinion.
Because of changes in traction, whether real or perceived, longer stopping distances are needed during the inclement weather. Typically this relates to drivers being more aware and cautious in their driving patterns. For a signalized intersection, the amount of amber time provided under dry conditions may not be sufficient to eliminate a potential dilemma zone in inclement conditions. Two factors contradict each other in determining the appropriate amber time. One is the decrease in speed, which reduces the needed amber time. The other is the reduction in deceleration rates, which causes the necessary amber time to increase. The following equation identifies the relationship in determining the appropriate amber time to eliminate dilemma zones.
where: tmin = minimum amber time d = perception-reaction time a = constant rate of braking deceleration (ft/s2) W = width of the intersection L = length of the car uo = approach speed
According to Garber and Hoel (1988) the deceleration rate under normal conditions is 27 percent the gravitational acceleration (0.27*32.2 ft/sec2 = 8.7 ft/sec2). With decreased traction in snowy conditions, drivers tend to be more comfortable with 20 percent the gravitational constant (6.4 ft/sec2). It is expected that for condition 7, snow packed roads, this deceleration rate may continue to decrease.
The 30 percent reduction in speed is not sufficient to account for the reduced deceleration and there is a resulting need for a 10 percent to 15 percent increase in amber time. For large intersections, more time is needed since the second term in the equation (clearing the intersection) is much greater with a wider intersection and yet slower speed with inclement weather. The needed increase typically is one-half to a whole second and therefore, is likely to have little effect on signal efficiency operations. It does, however, affect the dilemma zone and may have a substantial positive effect on reducing accidents. In addition to a reduced capacity for permitted left turn, the two sneakers are slower in clearing the intersection. Based on 136 observations during weather conditions 4 through 6, the sneakers took an average of 0.75 seconds longer to clear the intersection. Based on these findings, a one second increase in all-red time is recommended.
Chapter 5. Discussion
In section 4.2, the saturation flows and resulting headways of this research were compared to similar field values found in three other reports from the Literature Review. The purpose of this data comparison was not to validate results of any of the other three reports, but rather as a means to check the accuracy of the SLC field values. The winter of 2000 was one of the warmest on record and resulted in significantly fewer major snowfalls than most previous winters (National Climate Data Center, 2000). Therefore, very little traffic data was available for conditions beyond severity 6 in Salt Lake City. Below is a table summarizing the frequency of data collected for each weather condition.
Data Points Collected by Severity category
| Severity | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| Saturation flow | 108 | 55 | 92 | 33 | 9 | 14 | 0 |
| Speed | 137 | 60 | 118 | 22 | 5 | 0 | 0 |
The point at which an inclement weather signal timing plan should be implemented is subjective. Although there may be ways to automate the decision process, more research would be required to make this a reality. Speed and saturation flow by severity saw the largest decrease between severity categories 3 and 4. Category 3 is defined as "wet and snowing," and category 4 is defined as "wet and slushy." The addition of slush to the roadway seemed to be the boundary where vehicles begin to see the largest decrease in performance. This is defined as the beginning point of inclement weather, or the point at which a modified signal timing plan becomes appropriate. The values for speeds and saturation flows that should be used in developing the inclement weather signal timing plan must be based on percent reductions found for the average reduction among severity categories 4-7. Since there was not enough data collected for speeds past severity category 5, that value should be used.
The question of when to implement the signal timing plan for the entire network is much more subjective and difficult to answer. This decision could be triggered by an automatic process whereby the saturation flow or speed is monitored and when thresholds are met, the inclement timing plans are implemented. It is recommend that the decision to implement the new signal timing plan be done by a trained operator or engineer.
As some of the literature suggests, some isolated intersections may have some need for inclement weather modifications. Intersections that may warrant more detailed consideration are those that have steeply graded approaches or approaches with dilemma zone issues.
There are some considerations in developing an automatic alarm system. Such a system could alert an operator of the possibility that an inclement weather signal timing plan may be effective. There are several ways this could be done. Since many of the corridors for which the plan will be developed are instrumented with speed detectors. These speed detectors could be used to alarm operators to take a closer look at the surface conditions. Although a previous study found that RWIS equipment data did not correlate well to road surface conditions (Maki 1999), this could be further developed and tested with the increasing capability of the equipment. The data could then be used in conjunction with the sensor data to trigger an alarm.
In this research, issues are presented and several recommendations are made for circumstances that warrant an inclement weather signal timing plan. The recommendations are based on collected data and other research on the topic. There are plans to verify the findings with modeling and simulation. Similar to Maki (1999), the necessary software has been procured along with data from past projects and present traffic counts and timing plans. This allows modeling of normal and inclement weather conditions on a local corridor. The results of the modeling will provide added value to this project and hopefully will validate the recommendations. The modeling results also will provide estimated benefits of using an inclement weather signal timing plan. Upon completion, the results of this modeling will be provided in a supplemental report.
In general, changing cycle lengths will not benefit inclement weather conditions. Although the reality is that there likely will be a different optimum cycle length, this is most likely due to changes in traffic volume and turning movement flows during inclement weather periods. This assumption is based on the observed volume reduction during the Minnesota research (Maki 1999). There is insufficient SLC data to provide a confident statement recommending a general reduced volume during inclement weather, since there are many other external factors that control volume trends, other than inclement weather. Therefore, unless specific inclement weather traffic flows are collected on each corridor, the cycle length is assumed to remain constant. Instead, the offsets, splits, and clearance interval times should be changed for the inclement weather plans.
It is of note that we are recommending the extension of amber and all-red times. The Anchorage study (Bernardin Lochmueller and Associates Inc.1995) stated that if they were to change the amber times, the amber time would be reduced because of lower speeds. This conclusion, however, is probably erroneous because the deceleration also is decreased due to lower traction. Because they decided that the amber times would probably be lower, they recommended against changing them, due to the liability of decreasing amber times. Our recommendations present no risk of increased liability because we are recommending an increase in amber time. The Minnesota Study (Maki 1999) did not offer any discussion on amber or all-red times.
Most traffic signal controllers only allow a single amber and all-red time, which is applied to all plans. These parameters are not easily changed along with split, cycle, and offset times. We maintain this recommendation, however, in anticipation of future ability to change these times.