Temporal and Spatial Variation in Non-motorized Traffic in Minneapolis: Some Preliminary Analyses Spencer Agnew, Jason Borah, Steve Hankey, Kristopher Hoff, Brad Utecht, Zhiyi Xu, Greg Lindsey Thanks to: Shaun Murphy, City of Minneapolis Tony Hull, Transit for Livable Communities (TLC) TechPlan Research Program, ITS Institute, HHH State Local Policy Program, Center for Transportation Studies
Introduction and Overview Motivation and need for study Research objectives Data and methods Some preliminary results Some observations and conclusions Examples of potential uses Next steps
Motivation and Need for Study Transportation managers lack data about use of bicycle and pedestrian facilities Quality of data is poor; priority for data is high (BTS 2000) Federal, state, & local governments and nonprofits spending billions on new facilities New DOT-EPA-HUD programs focus on livability Need information & tools to plan, manage, evaluate, and optimize investments in facilities
Findings from Past Research: Non-motorized Traffic can be Modeled Traffic on multiuse trails in Indianapolis varies by season, day-of-week, hour-of-day, and location Daily traffic is correlated with weather, neighborhood demographics, urban form, and facility characteristics
Research Objectives Produce information to help transportation planners and managers work more effectively Assemble, analyze, and describe existing cycling and pedestrian counts in the Twin Cities metro Design system for using infrared monitors and other technologies to count traffic on urban multiuse trails and other facilities in region Begin collection of data to develop and validate models for forecasting urban trail traffic Test hypotheses related to the stability of measures of trail traffic and correlates of trail traffic
Data and Methods: Traffic Counts Non-motorized Traffic Counts in Minneapolis, Minnesota Method of observation Traffic observed Locations in Minneapolis Period of observation Number of observations Length of observations Manual Cyclist - separate Pedestrian - separate On and off-street bike facilities and no bike facilities (n=240) Magnetic Loop Detector Cyclist - separate Midtown Greenway (n=3) Infrared Cyclist and pedestrian combined Midtown Greenway (n=1) 2007-2009 2007-2009 12/09 1/10 458 + 2,500 54 (79% of possible days) 12-hour (n=43) 24 hours 24 hours 2-hour peak period (n=352) Other Limitations Human error Never validated Systematic undercount
Data and Methods: Correlates of Traffic Type of bike facility: Minneapolis DPW Street/road classification: Metropolitan Council Bus lines: Metro Transit Daily weather: National Oceanic and Atmospheric Administration (NOAA) Neighborhood demographics: U.S. Bureau of the Census Neighborhood land use: City of Minneapolis
Data and Methods: Approach Assembly and clean data (e.g., match locations and counts; map locations) Compute descriptive statistics (e.g., counts by facility type, presence of bus line) Compute planning ratios/scaling factors for extrapolating counts (e.g., time of day, day of week) Estimate 12-hour daily counts from short counts Model pedestrian and cycling traffic Assess need for validation of observation methods
Data and Methods: Location Attributes 240 locations (manual counts, bikes and peds) 100% of pedestrian count locations on sidewalks or trails 69% of bike count locations on streets with no bike facility 31% of bike count locations on bike facilities 18% are on-street facilities (i.e., bike lanes or shared traffic lanes) 13% are off-street facilities (e.g., trail) 61% of bike and ped count locations served by bus lines Number of repeat observations at locations varies
Type of Street / Facility Principal Arterial A-Minor B-Minor Collector Data and Methods: Location Attributes Daily Traffic Volume Number of count locations % all count locations 15,000-100,000 3 1% 0% 5,000-30,000 80 33% 18% 5,000-30,000 37 15% 43% % of count location type with bike facilities 1,000-15,000 49 20% 20% Local < 1,000 44 18% 18% Offstreet trail 0 27 11% 100 % All facilities 0-100,000 240 100% 31%
12-hour Bike Traffic Volumes (Actual observations (6:30 a.m. 6:30 p.m.; n=43)) Number of 12 hour observations Maximum traffic volume Mean traffic volume Median traffic volume Minimum traffic volume Average hourly traffic volume Off- Street (Trails) On- Street Bike Lane Share d Lane None All 6 10 1 26 43 1,005 1,157 71 901 1,157 584 625 71 215 358 642 541 71 202 247 229 240 71 13 13 49 52 6 18 30
Time of Day: 30-minute Bike Volumes (n=43)
Hourly Scaling Factors for Estimating 12-hour Volumes (n=43) Example: Multiplying 5-6 pm traffic by 8 yields 12-hour traffic volume.
Estimated 12-hour Volumes Highly Correlated with Actual 12- hour Volumes (n=43)
12-hour Bike Traffic Volumes (Actual & Estimated observations (6:30 a.m. 6:30 p.m.; Number of 12 hour observations Maximum traffic volume Mean traffic volume Off- Street (Trails) n=458) On- Street Bike Lane Share d Lane None All 97 81 5 275 458 6,701 3,138 964 3,39 4 6,701 851 566 450 362 502 Median traffic volume Minimum traffic volume Average hourly traffic volume 789 301 395 221 269 20 41 71 0 0 71 47 38 30 42
Mean Bike Traffic Volumes by Street & Facility Type (Actual & Estimated 12-hour observations (6:30 a.m. 6:30 p.m.; n=458) 1200 1000 Mean daily traffic volume 800 600 400 Bike facility No bike facility 200 0 Principal Arterial A-Minor B-Minor Collector Local Off-street Street Functional Type
Pedestrian Traffic Volumes by Street Type (Actual & Estimated 12-hour observations (6:30 a.m. 6:30 p.m.; Observation s Maximum volume Mean volume Median volume Minimum volume Average hourly volume Princip al Arterial A- Minor n=453) B- Mino r Collector Local All Street s Trails 6 161 75 58 63 363 90 150 18,15 3 6,23 0 13,424 1,476 18,15 3 3,150 87 1,091 903 1,447 355 832 295 86 676 274 461 230 331 122 36 0 19 4 0 0 0 7 91 75 121 30 69 25
Pedestrian Volumes, Bus Lines, & Trails (Actual & Estimated 12-hour observations (6:30 a.m. 6:30 p.m.; n=453) On Bus Route None Trails Observations 266 97 90 Maximum volume 18,153 8,492 3,150 Mean volume 1,123 531 295 Median volume 554 229 122 Minimum volume 0 0 0 Average hourly volume 94 44 25
Daily Bike Traffic Volumes on Greenway
Monthly Scaling Factors (Monthly Greenway Bike Traffic Relative to December Traffic) Example: July traffic is 10 to 30 times December traffic depending on year and location
Monthly Traffic Ratios for Greenways in Minneapolis and Indianapolis 20 18 16 Minneapolis mean monthly traffic ratios (cyclists) Indianapolis mean monthly traffic ratios (cyclists and peds) 14 12 10 Minneapolis traffic ratios show greater seasonality. 8 6 4 2 0 January February March April May June July August Septemb er October Novemb er Decemb er Differences could be associated with differences in counts (cyclists vs. cyclists & peds), characteristics of trails, or cultural or geographic factors.
Weekend-Weekday Greenway Bike Traffic Ratios (monthly mean daily traffic) Ratios > 1 indicate mean weekend traffic > mean weekday traffic
Need to Validate Automated Counters Infrared counter registering fewer events than magnetic loop detector Daily Traffic Count 450 400 350 300 250 200 150 100 50 - Hennepin Ave. Counter Site (Dec 2009 & Jan 2010) loop detector infrared Date
Variation in Time of Day Traffic by Mode and Facility Type
Explaining Variation in Traffic 12-hour traffic volume = f( Daily weather (temperature, deviation from average temperature, precipitation, wind speed) Neighborhood characteristics and form (population age, income, education, race, population density, land use mix) Traffic infrastructure type (street type, presence of bike facility, type of bike facility) Other factors (variables to be added)
Preliminary, Exploratory Independent Weather Variables Correlations Bicycle Traffic* Pedestrian Traffic* Maximum daily temperature Increases traffic Increases traffic Deviation from average temperature Not significant Not significant Precipitation (any) Decreases traffic Not significant Wind speed (average) Not significant Not significant *Bold, italicized significant at 10% level
Preliminary, Exploratory Independent Neighborhood Variables* Correlations Bicycle Traffic** Pedestrian Traffic** % population > 65, < 12 Increases traffic Not significant Median household income Decreases traffic Decreases traffic % population with college degree Increases traffic Increases traffic % Black population Decreases traffic Decreases traffic % other race Not significant Not significant Population density Not significant Not significant Land use mix Increases traffic Increases traffic *Estimated for Census block group where counting location occurs **Bold, italicized significant at 10% level
Preliminary, Exploratory Independent Infrastructure Variables Correlations Bicycle Traffic* (relative to local street, no bike facility) Pedestrian Traffic* (relative to off-street trail) Principal arterial Not yet modeled Not yet modeled A minor with bike facility Increases traffic (not applicable) B minor with bike facility Increases traffic (not applicable) Collector with bike facility Varies with model (not applicable) Local with bike facility Varies with model (not applicable) A minor Increases traffic Increases traffic B minor Increases traffic Increases traffic Collector Varies with model Increases traffic Local (base case) Increases traffic Off-street bike facility Increases traffic (base case) Presence of bus line Varies with model Increases traffic *Bold, italicized significant at 10% level
Preliminary Observations Non-motorized traffic follows recognizable patterns Bike volumes highest on off-street facilities and correlated with both street functional class and presence of bike facility Pedestrian volumes correlated with street functional class and presence of bus lines Measures of peak hour traffic volumes highly correlated with 12-hour volumes Time-of-day bike traffic on off-street facilities is similar to streets with/without facilities but peak hour traffic is higher % Time-of-day pedestrian traffic differs from bicycle traffic; includes greater noon-hour volumes (as % of 12-hour) Traffic varies in expected ways with weather and in complex ways with neighborhood characteristics and form
Examples of Potential Uses of Data Informing planning decisions Neighbors want to know how much bike traffic will increase on local street if bike lane added Mean 12-hour traffic, local street with facility (n=11): 387 Mean 12-hour traffic, local street, no facility (n=52): 277 Presence of bike lane: 40% greater traffic, on average 100 events / 12 hour day = 8 more bikes / hour; one more bike / 7 minutes Scaling factors and planning ratios enable broader application of limited data (e.g., estimates of 12-hour volumes; estimates of monthly traffic where no data exist Regression model will enable estimates of traffic where facilities might be added
Next Steps Complete analyses of archival data Refine statistical models (better specification of variables) Develop map of estimated volumes (web page?) Validate automated traffic counters Install new counters Publish reports