Short-duration Counts Research Articles

Evaluating the Potential of Crowdsourced Data to Estimate Network-Wide Bicycle Volumes

This research integrated and evaluated emerging user data sources (Strava Metro, StreetLight, and hybrid docked/dockless bike share) of bicycle activity data with conventional “static” demand determinants (land use, built environment, sociodemographics) and measures (permanent and short-duration counts) to estimate annual average daily bicycle traffic (AADBT). We selected six locations (Boulder, Charlotte, Dallas, Portland, Bend, and Eugene) covering varied urban and suburban contexts and specified three sets of Poisson regression models: city-specific models, an Oregon pooled model, and all cities pooled. Static variables, Strava, and StreetLight complemented one another, with each additional data source tending to improve the model performance. Sites with lower volumes were more difficult to predict, with considerable error in even the best-performing models. City-specific models in general exhibited improved fit and prediction performance. Expected prediction error increased by a factor of about 1.4 when using Strava or StreetLight alone, but without static adjustment variables, to predict AADBT. Combining Strava plus StreetLight, but without static variables, increased error slightly less; by 1.3 times. We also found that transferring the model specifications from one year to the next without re-estimating the model parameters resulted in a 10% to 50% increase in error rates across models, so such transfer is not recommended. The findings from this study indicate that rather than replacing conventional bike data sources and count programs, old “small” data sources will likely be very important for big data sources like Strava and StreetLight to achieve their potential for predicting AADBT.

Comparing avian species richness estimates from structured and semi-structured citizen science data

Citizen science, including structured and semi-structured forms, has become a powerful tool to collect biodiversity data. However, semi-structured citizen science data have been criticized for higher variability in quality, including less information to adjust for imperfect detection and uneven duration that bias the estimates of species richness. Species richness estimators may quantify bias in estimates. Here, we test the effectiveness of Chao1 estimator in eBird (semi-structured) by comparing it to averaged species richness in Breeding Bird Survey Taiwan, BBS (structured) and quantifying bias. We then fit a power function to compare bias while controlling for differences in count duration. The Chao1 estimator increased the species richness estimates of eBird data from 56 to 69% of the average observed BBS and from 47 to 59% of the average estimated BBS. Effects of incomplete short duration samples and variability in detectability skills of observers can lead to biased estimates. Using the Chao1 estimator improved estimates of species richness from semi-structured and structured data, but the strong effect of singleton species on bias, especially in short duration counts, should be evaluated in advance to reduce the uncertainty of estimation processes.

Data-Driven Approach to Quantify and Reduce Error Associated with Assigning Short Duration Counts to Traffic Pattern Groups

Traffic monitoring agencies collect traffic data samples to estimate annual average daily traffic (AADT) at short duration count sites. The steps to estimate AADT from sample data introduce error that manifests as uncertainty in the AADT statistic and its applications. Past research suggests that the assignment of a short duration count site to a traffic pattern group (TPG), characterized by known traffic periodicities, represents a significant but poorly quantified source of error. This paper presents an approach to quantify the range of errors arising from such assignments and to mitigate these errors using a novel data-driven assignment method. The approach uses simulated 48-hour short duration counts sampled from continuous count sites with known AADT to develop a benchmark of the total error expected when AADT is estimated from such samples. Likewise, the analysis produces a set of AADT estimates using temporal factors from pre-defined TPGs to quantify the range of assignment errors. The data-driven assignment method aims to mitigate these errors by minimizing the absolute mean deviation in AADT estimates produced from multiple short duration counts in a single year. The approach is applied to traffic data collected in Manitoba, Canada, as a case study. The results indicate that the mean absolute error from 48-hour short duration counts is 6.40% of the true AADT and that improper assignment can lead to a range in mean absolute errors of 9%. When applied to previously unassigned sites, the data-driven assignment method reduced mean absolute errors from 10.32%, using a conventional assignment method, to 7.86%.

Minimizing Annual Average Daily Nonmotorized Traffic Estimation Errors: How Many Counters Are Needed per Factor Group?

Accurate estimates of bicycle and pedestrian volume inform safety studies, trend monitoring, and infrastructure improvements. The Federal Highway Administration’s Traffic Monitoring Guide advises current practice for estimation of nonmotorized traffic. While methodologies have been developed to minimize error in estimation of annual average daily nonmotorized traffic (AADNT), challenges persist. This study provides new guidance for monitoring and volume estimation of nonmotorized traffic. Using continuous count data from 102 sites across six cities, the findings confirm that mean absolute percent error (MAPE) in estimated AADNT is minimized when seven-day short duration counts are collected in June through September and for 24-h counts, when data are collected Tuesdays through Thursdays (except for pedestrian-only counts). MAPE across all days (except holidays) and seasons was 34% for 24-h and 20–22% for seven-day short duration counts. The magnitude of bicycle and pedestrian volumes did not significantly affect estimation errors. For factor groups larger than one counter, the length of short duration samples may influence accuracy of AADNT estimates more than the number of counters per group, all else equal. To maximize precision of estimates of AADNT, four or more counters per factor group for bicycle and five or more for pedestrian travel monitoring are recommended. These findings provide guidance for practitioners seeking to establish or improve nonmotorized traffic monitoring programs.

Estimating Annual Average Daily Bicycle Traffic without Permanent Counter Stations

Traffic volumes are a basic unit of measurement for understanding the transportation system. As investments in bicycle infrastructure are made, similar measures are necessary for understanding this non-motorized mode of travel. Methods for estimating annual average daily bicycle traffic (AADBT) are still developing, but generally employ techniques used in the motorized traffic monitoring field whereby data from permanent counters are used to construct expansion factors that are then applied to short-duration counts. This approach requires a network of permanent counters and knowledge about how to group factors into appropriate categories based on patterns observed in both the short-duration and permanent counter data. The methods presented in this paper advance a new approach to estimating AADBT solely using short-duration counts. The Seasonal Adjustment Regression Model uses statistical models that relate the daily bicycle volume to daily conditions and weather variables at a given count location. These models are then used to predict daily volumes for the remaining days of the year. To verify this approach and determine the resulting error, levels of available short-duration counts using varying amounts of permanent count data were simulated. This method was then applied to short-duration bicycle counts from Eugene, Oregon. With sufficient short-duration count data, this method can produce AADBT estimates with minimal error and without requiring a network of permanent counters. This approach also circumvents the need to determine which expansion factors should be applied to different short-term count locations by using statistical models in place of expansion factors.

Annual Average Nonmotorized Traffic Estimates from Manual Counts: Quantifying Error

Across the United States, jurisdictions are investing more in bicycle and pedestrian infrastructure, which can benefit from nonmotorized traffic volume data. The design of nonmotorized counting programs varies. Whereas some agencies use automated counters to collect continuous and short duration counts, the most common type of bicycle and pedestrian counting is manual counting either in the field or from video. The objective of this research is to identify the optimal times of day to conduct manual counts for the purposes of accurately estimating annual average daily nonmotorized traffic (AADNT). This study used continuous bicycle and pedestrian counts from six U.S. cities to estimate AADNT and analyze estimation errors for multiple short duration count scenarios. Using two permanent counters per factor group reduces error substantially (> 50%); afternoon counts seem to be best for reducing error (2:00 to 6:00 p.m.). Error on Sunday is often as good as, if not better than, Saturday, contrary to what others have found. Arlington has the lowest AADNT estimation error (mean absolute percentage error), probably because of better data quality and higher nonmotorized traffic volumes, and Mount Vernon, Washington has the highest. Average AADNT estimation errors for the studied short duration count scenarios range from 30% to 50%. Error is lower for the commute factor group, bicycle-only counts, scenarios in which more peak hours are counted, and when more than one permanent counter is available to estimate adjustment factors.

Incorporating the Effect of Special Events into Continuous Count Site Selection for Pedestrian Traffic

This paper presents results from pedestrian monitoring research conducted in a dense urban environment in Winnipeg, Canada. Pedestrian counts were conducted in downtown Winnipeg using infrared pedestrian counters. Count sites were assigned to traffic pattern groups (TPGs) based on their response to special events occurring in the study area. Once these groups were established, eight continuous count sites were installed to initiate an ongoing pedestrian traffic monitoring program for the city. Traffic monitoring efforts have primarily focused on motorized travel. As more jurisdictions prioritize active transportation, addressing the need for network-level pedestrian data is essential to optimize engineering decisions. The first step to developing any system-wide traffic monitoring program is to define TPGs. These groups enable the spatial variation of short-duration counts to be adjusted to annual statistics by the temporal variation of similarly behaving continuous counts. Short-duration count sites were characterized by daily and hourly trends consistent with existing pedestrian traffic monitoring practices. Recognizing the influence of large evening events on pedestrian traffic, a metric was developed called the evening proportion ratio (EPR) to quantify the effect of special events. Based on the spatial distribution of EPR values, two TPGs were developed for downtown Winnipeg. These are the “urban utilitarian” and “urban utilitarian – event” groups. These groups were used to select continuous count locations for ongoing pedestrian traffic data collection. The importance of this research lies in its future applicability to other jurisdictions in developing a standard approach for urban transportation authorities to strategically implement pedestrian traffic monitoring programs.

Creating a Pedestrian Count Program in Downtown Winnipeg ( breakout presentation )

Background In the wake of an increasing number of North American jurisdictions moving to prioritize active modes of travel, research into traffic data collection has progressed beyond its infancy to a point where cities have begun network wide active transportation monitoring programs. This trend motivated by shifting societal values towards environmental sustainability, public health, and urban vibrancy; requires accurate and accessible data to best inform investment decisions and network improvements. Most pedestrian data collection of this variety has been conducted in an ad hoc manner on multi-use paths or simple sidewalk networks and often as a supplement to existing motorized traffic monitoring. In light of this, there is a clear need for a standard approach to collecting and reporting pedestrian volume data in highly urbanized areas. Method Documenting work performed for the City of Winnipeg, this research will provide a summary of actions taken and lessons learned in producing annual pedestrian volume statistics from Winnipeg’s downtown area. Using automated pedestrian counting devices, fourteen-day counts at 45 unique locations were conducted throughout the downtown area to characterize pedestrian activity. Following this, eight locations were selected to be monitored continuously throughout 2016. This enabled a detailed analysis of pedestrian trends throughout the year as well as the opportunity to extrapolate the results of existing short duration counts. Aim This data is currently being used to develop an ongoing pedestrian traffic-monitoring program for the City of Winnipeg. Once complete, this program will be used to estimate annual pedestrian volumes at locations where short duration counts are performed. This will serve as a vital tool to provide insights for transportation professionals, city planners, as well as community and activist groups to support design decisions. Results/Conclusion The two studies performed resulted in over 4.2 million pedestrians counted in 3480, 24-hour counts. This data has been used to distinguish two types of pedestrian activity in Winnipeg’s downtown area which can be used to estimate annual statistics. Additionally, summaries have been produced and presented to the City of Winnipeg, which describe yearly trends in pedestrian activity that will be used for engineering, planning, and policy design purposes.

Methodology to Characterize Ideal Short-Term Counting Conditions and Improve AADT Estimation Accuracy Using a Regression-Based Correcting Function

Transportation agencies’ motor vehicle count programs tend to be well established and robust with clear guidelines to collect short-term count data, to analyze data, develop annual average daily traffic (AADT) adjustment factors, and to estimate AADT volumes. In contrast, bicycle and pedestrian traffic monitoring is an area of work for most transportation agencies. In most agencies, there are a low numbers of counting sites and limited agency experience to manage a city-wide or state-wide system of collecting, processing, and using nonmotorized data. Short duration counts are used to estimate longer duration volumes such as AADT. Because bicycle or pedestrian short-term counts vary dramatically over time and significantly more than motorized vehicle counts, the direct application of motorized vehicle AADT estimation methods may be inadequate. The goal of this paper is to present a methodology that will enhance, if needed, existing AADT estimation methods widely employed for motorized vehicle counts. The proposed methodology is based on the analysis of AADT estimation errors using regression models to estimate a correcting function that accounts for weather and activity factors. The methodology can be applied to any type of traffic with high volume variability but in this research is applied to a permanent bicycle counting station in Portland, Oregon. The results indicate that the proposed methodology is simple and useful for finding ideal short-term counting conditions and improving AADT estimation accuracy.

Day-of-Year Scaling Factors and Design Considerations for Nonmotorized Traffic Monitoring Programs

General procedures, including the estimation of annual average daily traffic (AADT) from short-duration counts, have not been established for nonmotorized traffic monitoring programs. Continuous counts of nonmotorized traffic were collected at six locations on the off-street trail network in Minneapolis, Minnesota, in 2011. A new approach for estimating AADT values from short-duration counts, the use of day-of-year factors, is demonstrated. Analyses of variability in count data can be used to design a monitoring program that uses both continuous and short-duration counts of nonmotorized traffic. Five core conclusions may be useful for developing nonmotorized monitoring programs: ( a) day-of-year scaling factors have smaller error than does the standard (day-of-week and month-of-year) method of AADT estimation, especially from short-duration counts (<1 week); ( b) extrapolation error decreases with short-duration-count length, with only marginal gains in accuracy for counts longer than 1 week; ( c) errors in estimating AADT values are lowest when short-duration counts are taken in summer (or spring, summer, and fall) months (April through October); ( d) the impact of sampling on consecutive (successive) versus nonconsecutive (separate) days on AADT estimation is minimal but may reduce labor requirements; and ( e) the design of a traffic monitoring program depends on the acceptable error, equipment availability, and monitoring period duration. Trade-offs in short-duration-count lengths and estimate accuracy will depend on resource constraints. Analysts can use day-of-year factors to improve the accuracy of AADT estimation. Analyses of variability in traffic counts can strengthen the design of monitoring programs.

A Comprehensive Analysis of the Association of Highway Traffic with Winter Weather Conditions

This paper is based on the past several years of research carried out by transportation research group at the University of Regina, Canada to understand the impact of cold and snow on traffic volume during winter months in Canada. A detailed investigation of highway traffic variations i.e. total traffic, passenger car and truck traffic with severity of cold, the amount of snow, and various combinations of cold and snow intensities is presented here. These investigations were conducted using hourly traffic data from 350 permanent traffic counter sites, 6 Weigh in Motion sites and weather data from 598 weather stations located in the province of Alberta, Canada, from 1995 to 2010. Multiple regression analysis is used in the modeling process. The model parameters include three sets of variables: the amount of snowfall as a quantitative variable, categorized cold as a dummy variable, and an interaction variable formed by the product of these two variables. The study results indicate that the association of highway traffic flow with cold and snow varies with day of week, hour of day, and severity of weather conditions. A reduction of 1% to 2% in total traffic volume for each centimeter of snowfall is observed when the mean temperature is above 0°C. For the days with zero precipitation, reductions in total traffic volume due to mild and severe cold are 1% and 31%, respectively. An additional reduction of 0.5% to 3% per centimeter of snowfall results when snowfall occurs during severe cold conditions. Traffic volumes decreased with increase in the severity of cold temperatures. During extremely cold weather (below -25°C), the average winter daily traffic volume was reduced by about 30%. Weekend traffic volumes were more susceptible to cold than weekday numbers for all types of highways. Commuter and regional commuter roads experienced the lowest variations with cold. The impact of cold was very high for recreational roads and moderate for rural, long distance roads. This study also shows a clear indication in the reduction in daily traffic volumes due to snow (reductions between 7% and 17% for each centimeter of snowfall were observed). When individual vehicle classes were analyzed, it is found that passenger cars are more vulnerable to adverse weather conditions than trucks. Trucks are not as greatly affected as passenger cars by adverse weather conditions. Interestingly, the modeling results for one of the study sites reveal that higher truck traffic volumes can result during heavy snowfall (or other adverse weather conditions) in winter months due to shifting of trucks from secondary highways with poor winter maintenance to primary highways. This is contradictory to observations from other similar studies in the literature. The traffic-weather models developed from the past studies have two majorapplications in transportation engineering. These models have been tested for: (1) imputation of missing traffic data during winter months and (2) estimation of annual average daily traffic (AADT) from short-term traffic counts undertaken during winter months. The imputation efficiency of the traffic-weather models was compared with the most efficient imputation methods used by highway agencies. A new factor approach, incorporating weather conditions, was proposed in this paper to estimate AADT from short duration counts taken during winter months. Therefore, the developed traffic-weather models can be recommended to highway agencies to impute missing traffic data and to estimate AADT from short duration counts taken during winter months.

A statistical practical methodology of statewide traffic pattern grouping and precision analysis

Traffic data in general and traffic volume in particular are collected to determine the use and performance of the roadway system. Due to budget limitations, traffic volume cannot be counted day by day for every roadway within the state. Therefore, the volume on roadways without automatic traffic recorders (ATRs) can be determined by taking portable short-duration counts and using adjustment factors to produce annual average daily traffic (AADT) at a specific location. This study presents a statistical practical methodology that develops traffic pattern groups (TPGs) by combining roadways with similar traffic characteristics such as volume, seasonal variation, and land use in Delaware, USA. Monthly seasonal adjustment factors and their coefficient of variance (FCV) are analyzed for each group. To meet the desired confidence level and precision intervals, the TPGs’ ATR inventory is examined such that the required sample size is determined by the critical month.