bill davidson | pb | 410-243-4601 | [email protected] dawn mckinstry | pb | 714-973-4880 |...
TRANSCRIPT
Bill Davidson | PB | 410-243-4601 | [email protected] McKinstry | PB | 714-973-4880 | [email protected] Dowell | PB | 305-514-3125 | [email protected]
Calibration of the Regional Mode Choice Models for Los Angeles and Miami for New Starts Forecasting
11th National Transportation Planning Applications ConferenceMay 6-10, 2007, Daytona Beach, Florida
Session 18: Taken for a Ride: Ridership and Transit Forecasting
Introduction
Experience in Los Angeles and Miami Calibration Process Lessons Learned from New Starts
Calibration Process
Shift in how Mode Choice Models are Calibrated Detailed Review of Inputs Use of On-Board Transit Survey – A NECESSITY
Los Angeles Model
Mode Choice Model
Los Angeles Model
Study Area
Los Angeles Model
Challenges Encountered/Resolved1. Unrealistic Trip Patterns and Trip Lengths2. Uncongested Highway Speeds3. Metrolink Fares4. Drive Egress
Los Angeles Model
Validation Results
Los Angeles Model
Challenges Encountered/Resolved1. Unrealistic Trip Patterns and Trip Lengths2. Uncongested Highway Speeds3. Metrolink Fares4. Drive Egress
Los Angeles Model
Daily Boardings Summary
Mode Estimated Daily
Boardings
Observed(*) Daily
Boardings
Absolute Difference
Percent Difference
All Buses 1,282,500 1,184,700 97,800 8.3
Urban Rail 192,100 208,300 -16,200 -7.8
Commuter Rail 32,400 34,300 -1,900 -5.4
Rail Line Estimated Daily
Boardings
Observed Daily
Boardings
Absolute Difference
Riverside County 3,100 4,200 -1,100
91 Line 2,700 1,500 1,200
Riverside Line combined with 91 Line
5,800 5,700 100
Miami Model
Challenges Encountered/Resolved Cross County Trips Not Represented Nest Structure Model Coefficients College Trip Patterns Same as Other School Trips
Miami Model
Study Area
Miami Model
Updated Nest Structure
Miami Model
Review of Model Coefficients
Previous Model Updated Model
Drive access weight across all the trip
purposes is identical to the in-vehicle time.
In most models these coefficients should reflect
values that are more onerous than in-vehicle
time.
Single wait time. Wait time split into two parts, first and second
wait.
Miami Model
Miami-Dade College Trip Patterns
0%
5%
10%
15%
20%
25%
30%
35%
40%
Zip Codes
Per
cen
t
Model Data (Production Trip Ends) Place of Residence
0%
5%
10%
15%
20%
25%
30%
35%
40%
1 3 5 6 7 8 9 10 12
Distance (miles)
Per
cen
t T
rip
s
Model Distribution Place of Residence Distribution
Miami Model
Validation Results
Trip Purpose Target Value Estimated Value
Ratio of Estimated over Target
Home-Based Work Total Trips 3,952,017 3,952,019 1.00 Auto Trips 3,820,728 3,820,912 1.00 Transit Trips 131,289 131,107 1.00 Home-Based Non-Work Total Trips 6,939,560 6,939,266 1.00 Auto Trips 6,812,507 6,811,512 1.00 Transit Trips 127,053 127,754 1.01 Non-Home Based Total Trips 4,325,235 4,325,239 1.00 Auto Trips 4,271,676 4,271,248 1.00 Transit Trips 53,559 53,991 1.01
Transit Mode Observed Estimated Absolute Difference
Percent Difference
Local Bus (1) 237,823 234,744 -3,079 -1% Express Bus (1) 2,202 2,169 -33 -1% Jitney 3,007 3,002 -2 0% Tri-Rail 4,985 5,110 124 2% Metrorail 54,182 52,167 -2,016 -4% Metromover 38,486 36,939 -1,548 -4% Total 340,685 334,129 -6,556 -2% (1) – Includes both Miami-Dade and Broward County lines.
Lessons Learned
Review Key Model Inputs Person Trip Patterns
Los Angeles – Too few trips to downtown Miami – College trips
Highway and Transit Travel Time Los Angeles – Highway Speeds
Up to Date Surveys Calibration Target Values Observed Trip Patterns Path Building Process Verification Observed Trip Matrices