The Effects of Weather on Urban Trail Use: A National Study Alireza Ermagun, Tracy Haden-Loh, Greg Lindsey May 3, 2016
Rails to Trails Conservancy Trail Modeling and Assessment Platform Objectives Monitor trail traffic at 45-50 locations in 12-13 cities in 7 climatic regions for > 1 year Describe patterns & variation in trail traffic Develop hourly, day-of-week, monthly, and day-of-year factors Develop models for estimating trail use Support trail development efforts
Paper Objectives 1. Describe monitoring results a. Annual average traffic by mode 2. Estimate weather models a. Cyclists b. Pedestrians c. Total (mixed-mode) trail traffic 3. Compare variation in effects of weather across climatic regions, by mode
Research Findings 1. AADB, AADP, AADTT vary over three orders of magnitude (< 100 to > 1,000) 2. Weather effects different for cyclists and pedestrians 3. Weather models differ by region 4. Models explain a. Most in very cold region (Duluth) b. Least in hot dry region (San Diego) An aside: confirming the obvious? Trying to answer how much
Dept. of Energy Climate Regions TMAP Monitoring Locations
Research Approach: Key Steps 1. Data cleaning (QA/QC) 2. Model specification: variable selection 3. Modeling a. Negative binomial models b. Cragg & Uhler's Pseudo R-Squared to assess goodness of fit
Some Challenges in QA/QC WOD Bon Air West 4000 3000 2000 1000 Bike 0 NDDDJJJJFFMMMAAAMMMMJJJJJJAAASSSOOONNNDDDJJJFFFMMMAAAMMMJJJJJJAAASSSOOONNNDDDJJJJFFMMMM Chula Vista Bayshore Bikeway NB & SB 4000 3500 3000 2500 2000 1500 1000 500 0 NDDDJJJJFFMMMAAAMMMMJJJJJJAAASSSOOONNNDDDJJJFFFMMMAAAMMMJJJJJJAAASSSOOONNNDDDJJJJFFMMMM Bike
Cycling is More Seasonal than Walking (but not always) Bikes Billings MT 120 100 80 60 40 20 0 N-13 J-14 D-14 J-15 J-16 A-16 Peds Billings MT 1000 800 600 400 200 0 N-13 J-14 D-14 J-15 J-16 A-16 San Diego 5000 4000 3000 2000 1000 0 N-13 J-14 D-14 J-15 J-16 A-16 San Diego 600 500 400 300 200 100 0 N-13 J-14 D-14 J-15 J-16 A-16
% of Annual Traffic 14% Seasonality in Bike Traffic Varies by Region 12% 10% 8% Hot Dry SD 6% 4% 2% 0% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
% of Annual Traffic 14% Seasonality in Bike Traffic Varies by Region 12% 10% 8% 6% Hot Dry SD Hot Humid FL 4% 2% 0% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
% of Annual Traffic 14% Seasonality in Bike Traffic Varies by Region 12% 10% 8% 6% Hot Dry SD Hot Humid FL Mixed Dry NM 4% 2% 0% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
% of Annual Traffic 14% Seasonality in Bike Traffic Varies by Region 12% 10% 8% 6% Mixed Humid VA Hot Dry SD Hot Humid FL Mixed Dry NM 4% 2% 0% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
% of Annual Traffic 16% Seasonality in Bike Traffic Varies by Region 14% 12% 10% 8% 6% Mixed Humid VA Hot Dry SD Hot Humid FL Mixed Dry NM Marine WA 4% 2% 0% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
% of Annual Traffic 18% Seasonality in Bike Traffic Varies by Region 16% 14% 12% 10% 8% 6% Mixed Humid VA Hot Dry SD Hot Humid FL Cold MN Mixed Dry NM Marine WA 4% 2% 0% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
% of Annual Traffic 25% Seasonality in Bike Traffic Varies by Region 20% 15% 10% Mixed Humid VA Very Cold MN Hot Dry SD Hot Humid FL Cold MN Mixed Dry NM Marine WA 5% 0% Winter < 1% - < 10% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
3.0 Regional MADB / AADB Ratios 2.5 2.0 1.5 1.0 Mixed Humid VA Very Cold MN Hot Dry SD Hot Humid FL Cold MN Mixed Dry NM Marine WA 0.5 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
3.0 MADB / AADB Ratios in Cold and Very Cold Regions 2.5 2.0 1.5 1.0 Very Cold MN Cold MN Cold MT Cold IN Cold ME 0.5 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Variables in Weather Models Dependent variable = daily traffic volume (bike, ped, mixed-mode) Daylight hours Temperature Ave, squared Precipitation Dummy, interaction terms, squared Dew point Wind speed Snow depth Season Winter, Fall, Spring Holiday Weekend
Effects of Weather Vary by Region: Explanatory Power of Models Varies Pseudo R-squared Value Region Bike Models Ped Models Mixed-mode Models Very cold 0.91 0.82 0.88 Cold 0.52 0.43 0.52 Marine 0.66 0.14 0.38 Mixed dry 0.68 0.08 0.48 Mixed humid 0.56 0.48 0.56 Hot dry 0.21 0.05 0.19 Hot humid 0.45 0.19 0.36 National (all sites) 0.38 0.20 0.32
Effects of Weather on Cycling Vary by Region Weather Variable Very Cold Cold Marine Mixed Dry Mixed Humid Hot Dry Hot Humid Aggregate Day Light + + + + + + + Winter - + + - - Fall + + - + Spring - - Holiday - + + + Weekend + + + + + Tavg + + + + + + + + Tavg*Tavg - - - - - - - Precipitation=0 + + + + + + + + Precipitation - - - - - - - - Precipitation*2 + + + + + + + + DewPoint - - - - - + + DewPoint*DewPoint + + + + - - Avg Speed - - - - - - - - Scow Depth=0 + + + + + + + Constant + + - + - + - + Cragg & Uhler's R2 0.91 0.52 0.66 0.68 0.56 0.21 0.45 0.38
Research Findings 1. AADB, AADP, AADTT vary over three orders of magnitude (< 100 to > 1,000) 2. Weather effects different for cyclists, pedestrians 3. Weather models differ by region 4. Models explain a. Most in very cold region (Duluth) b. Least in hot dry region (San Diego)
Limitations - Lots 1. Limited number of sites in regions 2. Limited distribution of sites within regions 3. Subjectivity in QA/QC 4. Challenges in variable specification 5. Not controlling for interaction effects of culture, land use that may moderate effects of weather
Next Steps Review model specification Estimate general demand models Questions? Thank you!
Extra Slides with Modeling Details
Effects of Weather on Cycling Vary: Summary Variable Day Light Season Holiday Direction, Significance of Effect Positive, significant for all regions but cold Season: direction of effect varies by region (e.g., winter positive or negative) Varies, not significant in all regions Weekend Positive, but not significant in all regions Tavg Positive in all regions Tavg*Tavg Negative in all regions but very cold Precipitation=0 Dry (no precipitation: positive in all regions) Dew Point Negative in all regions but hot humid Ave wind speed Negative in all regions Snow Depth=0 No snow: positive (never snow in hot dry) Cragg & Uhler's R2 Range: 0.21-0.91
Effects of Weather on Walking Vary by Region Variables Very Cold Cold Marine Mixed Dry Mixed Humid Hot Dry Hot Humid Aggregate Day Light + + + + + + Winter + - - Fall - + + - + Spring + - Holiday + + + + Weekend + + + - + + + + Tavg + + + + + + + + Tavg*Tavg - - - - - - - Precepitation=0 + + + + + + + + Precipitation - - - - - - - Precipitation*2 + + + + DewPoint - - - - - + + DewPoint*DewPoint - + - - Avg Speed - - - - - + - Scow Depth=0 + + + Constant + + + + + - - + Cragg & Uhler's R2 0.82 0.43 0.14 0.08 0.48 0.05 0.19 0.2
Effects of Weather on Walking: Summary Variable Day Light Season Holiday Weekend Tavg Tavg*Tavg Direction, Significance of Effect Positive, significant for all regions but very cold, hot dry Season: direction of effect varies region (e.g., winter positive or negative) Varies: postive in some regions, insignificant in others Positive, significant in all regions but mixed dry (negative) Positive in all regions Negative in all regions but very cold Precepitation=0 Dry (no precipitation: positive in all regions) DewPoint Negative in all regions but hot humid Avg Wind Speed Negative in all regions but hot humid Snow Depth=0 No snow: varies, but positive in some regions Cragg & Uhler's Range: 0.05-0.82 R2