Non-Motorized Traffic Exploratory Analysis

Similar documents
Understanding Land Use and Walk Behavior in Utah

2014 Data Collection Project ITE Western District

Project Appraisal Guidelines

WOODRUFF ROAD CORRIDOR ORIGIN-DESTINATION ANALYSIS

Technical Memorandum #2 Future Conditions

Optimization of Short-Term Traffic Count Plan to Improve AADT Estimation Error

MADISON, WI STONE HOUSE DEVELOPMENT 1000 E. WASHINGTON AVENUE REDEVELOPMENT TRANSPORTATION STUDY DECEMBER 14, 2015

Neighborhood Locations and Amenities

Trip Generation Study: A 7-Eleven Gas Station with a Convenience Store Land Use Code: 945

Trip and Parking Generation Study of Orem Fitness Center-Abstract

Traffic Impact Study

Susan Clark NRS 509 Nov. 29, 2005

MEMORANDUM. The study area of the analysis was discussed with City staff and includes the following intersections:

Southwest Light Rail Transit Bicycle Facility Assessment Technical Memorandum #3 Prioritization. July 23, P age

California Urban Infill Trip Generation Study. Jim Daisa, P.E.

Complete Bicycle Network Project Prioritization

Appendix C Final Methods and Assumptions for Forecasting Traffic Volumes

Mapping Accessibility Over Time

Appendix BAL Baltimore, Maryland 2003 Annual Report on Freeway Mobility and Reliability

Committee Meeting November 6, 2018

Analysis and Design of Urban Transportation Network for Pyi Gyi Ta Gon Township PHOO PWINT ZAN 1, DR. NILAR AYE 2

FHWA Planning Data Resources: Census Data Planning Products (CTPP) HEPGIS Interactive Mapping Portal

Figure 8.2a Variation of suburban character, transit access and pedestrian accessibility by TAZ label in the study area

FHWA GIS Outreach Activities. Loveland, Colorado April 17, 2012

GIS Analysis of Crenshaw/LAX Line

Speed Limit Review. Montague Road, West End. Prepared for Brisbane City Council CEB06842 CEB06842

City of Hermosa Beach Beach Access and Parking Study. Submitted by. 600 Wilshire Blvd., Suite 1050 Los Angeles, CA

GIS Technology and Tools for Long Range Transportation Planning in the National Park Service

3.0 ANALYSIS OF FUTURE TRANSPORTATION NEEDS

Parking Study MAIN ST

3/21/2019. Q: What is this? What is in this area of the country? Q: So what is this? GEOG 3100 Next Week

DEVELOPING DECISION SUPPORT TOOLS FOR THE IMPLEMENTATION OF BICYCLE AND PEDESTRIAN SAFETY STRATEGIES

The Effects of Weather on Urban Trail Use: A National Study

BROOKINGS May

RECORD OF MEETING. Region Five Development Commission

CLAREMONT MASTER PLAN 2017: LAND USE COMMUNITY INPUT

NATHAN HALE HIGH SCHOOL PARKING AND TRAFFIC ANALYSIS. Table of Contents

WEBER ROAD RESIDENTIAL DEVELOPMENT Single Family Residential Project

HIRES 2017 Syllabus. Instructors:

Transit Time Shed Analyzing Accessibility to Employment and Services

HORIZON 2030: Land Use & Transportation November 2005

A Comparison of the Accessibility of Three Neighborhoods Institutions and Amenities in Frederick, MD

South Western Region Travel Time Monitoring Program Congestion Management Process Spring 2008 Report

Appendix B. Land Use and Traffic Modeling Documentation

Down East RPO. Strategic Prioritization Office of Transportation Local Input Point Assignment Methodology. Introduction

PALS: Neighborhood Identification, City of Frederick, Maryland. David Boston Razia Choudhry Chris Davis Under the supervision of Chao Liu

FDOT Level 2 Roundabout b/c Evaluation

Keywords: Air Quality, Environmental Justice, Vehicle Emissions, Public Health, Monitoring Network

NSP Electric - Minnesota Annual Report Peak Demand and Annual Electric Consumption Forecast

Palmerston North Area Traffic Model

Impact of Day-to-Day Variability of Peak Hour Volumes on Signalized Intersection Performance

2012 FLOODS AND DEBRIS REMOVAL

CALOTS Upgrade for Performance Monitoring

StanCOG Transportation Model Program. General Summary

RE: Existing and Future Parking Demand Analysis St. Joseph Center Expansion

ADVENTURES IN THE FLIPPED CLASSROOM FOR INTRODUCTORY

Proposed Scope of Work Village of Farmingdale Downtown Farmingdale BOA Step 2 BOA Nomination Study / Draft Generic Environmental Impact Statement

Encapsulating Urban Traffic Rhythms into Road Networks

CITY OF NEW LONDON WINTER ROAD & SIDEWALK MAINTENANCE POLICY

Annual Collision Report

Advancing Transportation Performance Management and Metrics with Census Data

Montmorency County Traffic Crash Data & Year Trends. Reporting Criteria

I. M. Schoeman North West University, South Africa. Abstract

ADAPTIVE SIGNAL CONTROL IV

Chao Liu, Ting Ma, and Sevgi Erdogan National Center for Smart Growth Research & Education (NCSG) University of Maryland, College Park

2129 NORTH MAIN STREET HOTE PROJECT ULI SHARED PARKING STUDY City of Santa Ana, California

Regional Performance Measures

Cost-Benefit Analysis of the Pooled- Fund Maintenance Decision Support System: Case Study

An Assessment of People, Place and Business on Syracuse s Near Northside

Regional Performance Measures

Douglas County/Carson City Travel Demand Model

The I-81 Challenge: Study update and overview of travel demand model CNY Engineering Expo November 12, 2012

MINNESOTA DEPARTMENT OF TRANSPORTATION Engineering Services Division Technical Memorandum No MAT-02 October 7, 2014

Assessing spatial distribution and variability of destinations in inner-city Sydney from travel diary and smartphone location data

Local Economic Activity Around Rapid Transit Stations

A Socioeconomic Analysis of the Spatial Distribution of Fire Hydrants. History of Portland Fire Hydrants

QUANTIFICATION OF THE NATURAL VARIATION IN TRAFFIC FLOW ON SELECTED NATIONAL ROADS IN SOUTH AFRICA

Using Tourism-Based Travel Demand Model to Estimate Traffic Volumes on Low-Volume Roads

H. R heading to fund the award and oversight by the Administrator of grants made under this heading. FEDERAL HIGHWAY ADMINISTRATION

Keywords: Travel Behavior; Transportation Finance; Global Positioning Systems (GPS)

How ACS Data Can Make Smart Cities Even Smarter: A method for combining bike sensor data with ACS demographic data Peter Viechnicki, Deloitte Center

2015 Grand Forks East Grand Forks TDM

5.1 Introduction. 5.2 Data Collection

Status Report: Ongoing review of O-D cellular data for the TPB modeled area

APPENDIX C-6 - TRAFFIC MODELING REPORT, SRF CONSULTING GROUP

Crow River Plaza - Retail Development South Diamond Lake Rd. Rogers, MN FOR LEASE 1,024 SF Retail Space. Lease Rate: $16.

NEW YORK DEPARTMENT OF SANITATION. Spatial Analysis of Complaints

Spatial Variation in Local Road Pedestrian and Bicycle Crashes

Market Street PDP. Nassau County, Florida. Transportation Impact Analysis. VHB/Vanasse Hangen Brustlin, Inc. Nassau County Growth Management

Bicycle and Pedestrian Count Program 2018 Annual Report (Year 5) July 2018

Monroe County: Key West and Lower Keys

Developing the Transit Demand Index (TDI) Gregory Newmark, Regional Transportation Authority Transport Chicago Presentation July 25, 2012

Hennepin GIS. Tree Planting Priority Areas - Analysis Methodology. GIS Services April 2018 GOAL:

TRAFFIC FORECAST METHODOLOGY

Focused Traffic Analysis for the One Lincoln Park Project

COUNCIL POLICY MANUAL

Winter Weather Safety Tips. From your friends at South Brunswick Township Department of Public Works

Expanding the GSATS Model Area into

How GIS Can Help With Tribal Safety Planning

Transcription:

Non-Motorized Traffic Exploratory Analysis --------------------------------------------------------------- Chao Wu Advisor: Professor Greg Lindsey Humphrey School of Public Affairs University of Minnesota April 28, 2014

Abstract In 2012, Minnesota Department of Transportation (MnDOT) implemented the Minnesota Bicycle and Pedestrian Counting Initiative, a non-motorized transportation (bicycle and pedestrian) count project. A voluntary statewide count was repeated in spring 2013, and fall 2013. Between September 10 th, 2013 and September 19 th, 2013, 14 communities conducted counts in their municipalities using a standard count form provided by MnDOT. Counting from September 2013 has been analyzed in the report. The first objective of the report is to summarize results of non-motorized traffic counts. In addition, the second objective of the report is to explore potential relationship between the non-motorized counts and several dependent variables as followed: Anural average daily traffic (AADT) counts Population density Median household income Education Unemployment rate City class

Introduction Nowadays, there are more and more pressures for government to encourage and foster both recreational and utilitarian non-motorized transportation. It becomes more and more important to understand non-motorized transportation patterns. Understanding how does bicycle and pedestrian volume react with the factors listed above is important for the planning and management of non-motorized transportation. Understand these relationships are important because it could prioritize the investments, and help to make more efficient investments. (Lonetta, Angela, Jenny, Jason, 2013). The following report is divided into five parts. First of all, the background part will introduce the data collecting method, and provide summary results by municipality. Followed by that, the methodology part will introduce how the assumptions have been made. Then, during the techniques part, different methods will be introduced. Followed by that, the results part will present the major findings of the research. In addition, conclusion and recommendation will be presented to discuss the possible outcomes. In the end, appendix A through X will summarize count for each municipality. Background In September 2013, 14 Minnesota communities completed 238.25 hours of short-term manual bicycle and pedestrian counts at 55 locations during peak weekday and peak weekend traffic hours, following MnDOT procedures. Participating communities

included International Falls, Carltons, Grand Rapids, La Prairie, Coleraine, St. Cloud, St. Joseph, Sartell, Sauk Rapids, Waite Park, Fargo, Moorhead, West Fargo, and Dilworth. However, Fargo, and West Fargo are excluded in the following analysis due to insufficient AADT information. Table 1 below summarizes community counts by city size, number of locations monitored, and total hours of counts. Minnesota divides their cities into four classes. Class 4 city has population of 10,000 or less, class 3 city has population from 10,000 to 20,000, class 2 city has population from 20,000 to 100,000, and class 1 city has population over 100,000. Since this report focuses on cities of Minnesota, so Fargo, and West Fargo are excluded during the analysis. Six communities, or 43%, counted in only one location. Also, MnDOT encouraged communities to count on Tuesdays, Wednesdays, and Thursdays in September during evening peak hours, especially from 4:00PM to 6:00PM. Counting on Saturdays and/or other times identified as providing value to local initiatives were also encouraged. In the end, last two columns showed average bike count and average pedestrian count for each municipality.

Table 1: Location and Hours Counted in Minnesota/North Dakota Municipalities Class Population Number of Total Mean Hourly Mean Hourly Locations Hours Bike Count Pedestrian Count International Falls 4 6424 3 14.25 8 16 Carltons 4 854 2 8 10 12 Grand Rapids 3 10916 4 40 3 5 La Prairie 4 663 1 10 2 1 Coleraine 4 1983 1 10 2 2 St. Cloud 2 65986 6 12 12 83 St. Joseph 4 6646 1 2 3 3 Sartell 3 16183 1 2 12 9 Sauk Rapids 3 12965 1 2 7 4 Waite Park 4 6680 1 2 3 5 Fargo 1 223490 20 80 12 31 Moorhead 2 39039 9 36 23 35 West Fargo 2 27478 3 12 6 19 Dilworth 4 4091 2 8 2 2 In addition, Appendix A through L showed a detailed count for each municipality. It included brief background information for each municipality, a summary of average bikes and pedestrian count of all sites in each municipality, a conclusion of maximum bikes and pedestrians count per hour of all sites in each municipality. Also, tables of each count location and graphs of each count location attached at end of each Appendix. Methodology for Exploratory Analysis Table 2 below included dependent variables, independent variables and assumption of how independent variables and dependent variables correlated with each other.

Table 2: Dependent Variables vs. Independent Variables Dependent variables Independent variables Assumption Bicycle counts Average Annual Daily Traffic Positively (AADT) correlated Pedestrian counts Population Density Positively correlated Median household income Positively correlated Education Positively correlated Unemployment rate Positively correlated Job Density Positively correlated City Class Positively correlated In terms of dependent variables, bicycle counts and pedestrian counts are targeted dependent variables. In terms of independent variables, there are Average Annual Daily Traffic (AADT), population density, median household income, education, unemployment rate, job density and city class. The assumption was that all independent variables were positively correlated with dependent variables. It means if there is an increase of independent variable, there is an increase of dependent variable. Techniques Generating Hourly Vehicular AADT The following part will discuss two different methods been used in the analysis. The method is to reverse extrapolate of AADT into a two-hour traffic count. Since the AADT

data from MnDOT website is the average daily count, but we only have a two hour count for bicycle and pedestrian. The initial thought was to extrapolate two-hour count into daily counts; then daily counts to weekly counts; finally to and annual daily traffic count. Since we have limited non-motorized adjustment factor, the new approach is to reverse extrapolate AADT into a 2-hour short duration count. MnDOT has provided an AADT factor for weekdays and weekends in September. With this factor, it is possible to reverse extrapolate AADT counts into September daily counts. In terms of hours of day factor, MnDOT provided a spreadsheet that has automatic traffic recorder (ATF) information (MnDOT, 2013). The ATF report has 230 randomly picked roads in Minnesota with every hour traffic count throughout 2013. Among all roads information, five random selected roads have been chosen. Then calculate the average hours of day factor. In this way, we will able to match motorized traffic with non-motorized traffic. Generating Census Related Variables The second methodology is to use 2010 census data on other dependent variables at block group level. Considering access distance between bicycles and pedestrians are different, 1 mile buffer or access zone was created for bicycles, and 1/2 mile buffer or access zone was created for pedestrians. Within the buffer zone, 2010 data for median household income, education level, employment distribution, and population were generated from Census Bureau. In order to create the buffer zones, GIS maps with road names are used from Census Bureau website. First of all, buffer the count location by 0.5 mile and 1 mile

respectively. Then clip the block group information into each buffer zone. In addition, information about median household income, education level, employment distribution, and population are also found from Census Bureau website. After clip the block group, manually match the above data coordinated with block group numbers. Results The following part will show several multiple regressions analysis. There are two sets of data are been analyzed. The first set of data has 19 observations without AADT. The second set of data has 11 observations with AADT. The reason is that MnDOT only provides AADT for limited numbers of roads. In addition, since participated communities do not have class one city, the multiple regressions use class two city and class three city as oppose to class four city. During the analysis, we want the absolute value of t-value to be greater than 1.95 or 2.58 to be significant at confidence level of 95% and 99% respectively, meanwhile the absolute value of p-value to be smaller than 0.05 or 0.01 to be significant at confidence level of 95% and 99%. Table 1 below showed the multiple regressions for 0.5 Mile pedestrian count. Based on t-value, there is no independent variable showing significance. Similarly, in terms of p-value, there is also no independent variable showing significance. Also, the adjusted R square is as low as -0.03, it indicates all independent variables combine explaining 3% of the dependent variables.

Table 1: 0.5 Mile Pedestrian Count Coefficients Standard Error t Stat dp-value Intercept 15.48 20.01 0.77 0.46 Population Density 0.00 0.00-1.03 0.32 Median income 0.00 0.00 0.37 0.72 College degree or higher -41.24 54.70-0.75 0.47 Unemployment rate -59.64 123.57-0.48 0.64 Jobs Density 0.00 0.00 1.41 0.19 Class II 3.49 9.96 0.35 0.73 Class III -0.62 6.72-0.09 0.93 Adjusted R Square -0.03 Observations 19.00 Table 2 below showed the multiple regressions for 0.5-mile bicycle count. There is no significance of t-value, due to all t-values are smaller than 1.96. Similarly, there is no significance of p-value, due to all p-values are greater than 0.05. On the other hand, the adjusted R square indicates all independent variables combine explaining 25.8% of the dependent variables. Table 2: 0.5 Mile Bike Count Coefficients Standard Error t Stat P-value Intercept -1.28 10.96-0.12 0.91 Population Density 0.00 0.00 0.57 0.58 Median income 0.00 0.00 0.96 0.36 College degree or higher -3.00 29.97-0.10 0.92 Unemployment rate -3.45 67.71-0.05 0.96 Jobs Density 0.00 0.00 0.95 0.36 Class II -4.95 5.46-0.91 0.38 Class III -2.84 3.68-0.77 0.46 Adjusted R Square -0.258 Observations 19.00

Table 3 below showed the multiple regressions for 1 Mile pedestrian count. According to the data, only unemployment rate is showing potential relationship with pedestrian count, due to higher t-value. However, in terms of p-value there is no dependent variables showing significance because all p-values are greater than 0.05. In terms of adjusted R square, it indicates only 16% of dependent variables can be explained by independent variables. Table 3: 1 Mile Pedestrian Count Coefficients Standard Error t Stat P-value Intercept 60.50 25.22 2.40 0.04 Population Density 0.00 0.01-0.39 0.71 Median income 0.00 0.00-0.23 0.83 College degree or higher -100.89 71.78-1.41 0.19 Unemployment rate -441.20 209.98-2.10 0.06 Jobs Density 0.00 0.00 0.69 0.50 Class II 6.80 12.98 0.52 0.61 Class III -1.25 6.48-0.19 0.85 Adjusted R Square 0.1607 Observations 19 Table 4 below showed the multiple regression for 1 Mile bicycle count. According to the data, job density and city class are showing higher t-value but not enough to be significant. In terms of p-value, there is no significance showing due to no smaller than 0.05 p-values. In terms of adjusted R square value, it indicates all independent variables combine explaining 14% of the dependent variables.

Table 4: 1Mile Bike Count Coefficients Standard Error t Stat P-value Intercept 0.12 14.56 0.01 0.99 Population Density 0.00 0.01-0.34 0.74 Median income 0.00 0.00 0.18 0.86 College degree or higher 31.16 41.45 0.75 0.47 Unemployment rate -71.67 121.25-0.59 0.57 Jobs Density 0.00 0.00 1.21 0.25 Class II -9.23 7.49-1.23 0.24 Class III -5.03 3.74-1.34 0.21 Adjusted R Square -0.141 Observations 19.00 Summary of Modeling Results According to the results above, there is no single dependent variable showing significant relationship with independent variables. However, it seems only population density; job density and city class are showing significance during the analysis. The following multiple regressions will only use the above dependent variables and exclude the rest. Table 5 below showed the multiple regressions for 0.5 Mile pedestrian count. According to the data, population density and job density are showing significant t-value. In terms of p-value, both population density and job density are showing a small p-value. In terms of adjusted R square value, it indicates all independent variables combine explaining 14.85% of the dependent variables. Even though population density and job density are showing significance, but it is difficult to determine whether those dependent variables correlate

with each other or they both correlate with pedestrian count. As can be seen in table 6, a correlation analysis is presented. According to the table, job density and population density are showing the highest correlation number, which is 0.79. It means denser the area is more jobs there will be. Also, population density and job density are also showing relative high correlation with city class. It indicates larger city is intent to have denser population and more jobs. However, all dependent variables are showing less correlation with pedestrian count. To sum up, the tables above reflect land use pattern more than anything else. There is no significant evidence showing pedestrian and bicycle counts are correlated with those dependent variables. Table 5: 0.5 Mile Pedestrian Count Redo Coefficients Standard Error t Stat P-value Intercept 7.62 3.33 2.29 0.04 Population Density -0.01 0.00-1.90 0.08 Job Density 0.00 0.00 2.15 0.05 Class II 0.99 6.28 0.16 0.88 Class III -2.98 4.89-0.61 0.55 Adjusted R Square 0.1485 Observations 19

Table 6: 0.5 Mile Pedestrian Count Redo Ped Count Population Density Jobs Density Class II Class III Ped Count 1.0000 Population Density 0.0500 1.0000 Jobs Density 0.3781 0.7954 1.0000 Class II 0.2689 0.6280 0.6979 1.0000 Class III -0.2511-0.2348-0.2911-0.4060 1.0000 Conclusion According to results above, there are no significant relationship between dependent variables and independent variables. There are several reasons causing those results: Too few observations Other independent variables o Time of day o Weather Randomness Location There are too few observations causing data extremely unstable. In addition, there are other independent variables. For example, different communities recorded on different time of day. Also, normally, we expect to see more bicycles and pedestrian during morning or evening. Similarly non-motorized traffic is heavily relying on the weather. More pedestrians and bikes are expected during good weather days. Furthermore, non-motorized traffic has a large proportion of randomness. For example, some bicycle

users occasionally use bikes to exercise. There is only a small portion of the users use bike as a transportation tool. Last but not least, location is one of the most important issues. For example, more pedestrians and bikes are expected in locations such as schools or trails. Similarly, almost no pedestrians or bikes are expected near highway intersections. In addition, Suggestion Based on the analysis above, even though the figures show a less relative correlation, a few recommendations are also provided. First of all, MnDOT should continue to encourage communities to participate in annual statewide bicycle and pedestrian counts. More counts will provide more adequate local non-motorized transportation patterns. Communities should be encouraged to complete several counts at each location in order to develop patterns for more thorough analysis. In addition, the analysis results would be better if each location is organized by the street type. Also, recording the weather is also important in order to avoid any useless data.

Reference 1. "Minnesota Department of Transportation." / MnDOT.gov. N.p., n.d. Web. 28 Apr. 2014. 2. Hanson, Lonetta, Angela Laird, Jenny Monson-Miller, and Jason Radde. "MnDOT Report." (2013): n. pag. Web. 3. "Welcome to an Engaged Community." International Falls, MN. N.p., n.d. Web. 28 Apr. 2014. 4. "City Of Carlton." City Of Carlton. N.p., n.d. Web. 28 Apr. 2014. 5. "Grand Rapids Herald-Review." Grand Rapids Herald-Review. N.p., n.d. Web. 28 Apr. 2014. 6. "City of Coleraine; Recreation, History & Attractions in Minnesota." City of Coleraine; Recreation, History & Attractions in Minnesota. N.p., n.d. Web. 28 Apr. 2014. 7. "Welcome to an Engaged Community." St. Cloud, MN. N.p., n.d. Web. 28 Apr. 2014. 8. "Sauk Rapids." Sauk Rapids. N.p., n.d. Web. 28 Apr. 2014. 9. "Welcome to an Engaged Community." St. Joseph, MN. N.p., n.d. Web. 28 Apr. 2014. 10. "City of Moorhead : Home." City of Moorhead : Home. N.p., n.d. Web. 28 Apr. 2014. 11. "Dilworth, MN." Dilworth, MN. N.p., n.d. Web. 28 Apr. 2014. 12. "Census Bureau Homepage." Census Bureau Homepage. N.p., n.d. Web. 26 Apr. 2014.