Spatial

1
Spatial Temporal Allocation of On-Road Emissions

• CCOS Technical Committee
• November 28, 2006

Prepared by Tom Kear, Ph.D., P.E. Dowling
Associates Debbie Niemeier, Ph.D., P.E. UC Davis
2
Presentation Overview

• Preview of key issues
• On-road proportion Prior CCOS work
• Major trends identified in the literature heavy
  duty modeling practice
• Critical assumptions
• Findings
• Phase II priority projects

3
Preview Of Key Issues

• The ITN used to develop the base-year (2000)
  inventory is not applicable to future years
• Heavy-duty vehicle activity, in general, is not
  being modeled, but is assigned to roads as a
  percentage of light duty vehicle activity
• Speed post-processing has been to shown
  dramatically affect emission estimates under
  certain conditions
• Current modeling techniques are not capturing the
  spatial distribution of weekend travel

4
On-Road Proportion Of Emissions

• On-Road contributes about 1/3 of the ROG
  inventory
• Diesel vehicles are not an important source of ROG

5
On-Road Proportion Of Emissions

• On-Road contributes about 50 of the NOx
  inventory
• Trucks account for about 3 of VMT but 30 of
  on-road NOx

6
Prior CCOS Work

• BURDEN 2002 emissions allocated to grid cells
  using DTIM4
• Integrated Transportation Network (ITN) from
  individual county (loaded) travel demand model
  networks
• Temporal allocations assigned per BURDEN and
  available traffic counts

7
Prior CCOS Work
8
Prior CCOS Work
9
CCOS NOx, TOG, HDV NOx
10
Critical assumptions

• CCOS assumes uniform growth of vehicle activity
  across regions
• Note the variation in growth forecasts, ranging
  from none to more than 10x (e.g., 1,000)
• ITN needs to be rebuilt using loaded networks for
  each analysis year (interpolated trip tables)
  prior to DTIM runs

11
Prominent Trends in Literature

• Light/heavy vehicle ratio differ by day of week
• Less truck activity on weekends, but the ratio of
  LDV/HDT increases
• Ratios vary by geographic location

• Weekdays (Mon-Thurs) have similar temporal
  allocation
• Saturday and Sunday are often very different from
  each other

12
Prominent Trends in Literature
Speed post processing has a significant effect on
congested emissions
13
Prominent Trends in Literature
Table 10. Annual unpaved road VMT in California
Harvest VMT Nonharvest VMT Total Statewide VMT
4,945,329 468,023,838 472,969,167

• Statewide HVMT accounts for only about 1 of the
  annual total. The low HVMT suggests that changes
  in harvest hauling traffic patterns will not
  dramatically affect emissions for a typical day.
  
• Current activity factor for nonagricultural
  unpaved roads underestimated vehicle activity for
  Forest and Woodland and Urban Residential areas,
  but overestimated vehicle activity in Grasslands,
  Sand dunes and Scrubland and Urban Interface
  areas.

14
Heavy Duty Vehicles

• Not modeled but captured during calibration by
  increasing non-home-based-trips to match counts
• True freight models aggregate trip tables from
  inter county commodity flow data and regional
  gravity models.
• Trucks not well captured by SJV phase II truck
  model, or any of the 8 RTPA models.
• 2025 SJV Phase III truck model forecast is being
  extrapolated from 1978 commodity flow surveys

15
Heavy Duty Vehicles
SJV Goods Movement Study Phase II (2004)
16
Critical assumptions

• The current approach assumes weekend and weekday
  trip distribution is identical, only the number
  of trips generated changes
• Just matching base year creates a forecasting
  problem because behavioral component is lacking
• Heavy duty vehicle activity is assumed to be
  distributed similarly to the light duty vehicle
  activity on all RTPA networks.
• Assumes that trip based emission factors are
  applicable to links
• Existing and future activity is assumed to follow
  the same spatial / temporal distributions

17
Findings from Phase 1

• Areas of uncertainty
• Spatial changes between weekday-weekend activity
• Where are the trucks?
• Spatial mismatch between activity data
  emissions rates
• Impact of better transportation data (refinement
  of spatial network, speed post processing, and
  the treatment of trip ends)
• Impact of seasonality on agricultural goods
  movement

18
Findings from Phase 1

• Best way to group daily hours of travel?
• Importance of speed post processing
• Trucks are not well represented
• Weekend activity is not well represented

19
Phase II priority projects
Task Description Cost
2010-2020 forecasts Statewide Model, DTIM, spline smoothing. 80 K
Improve truck data Model truck activity on highways and arterials, integrate w/ task (1) 75 K (115 K if counts needed)
Speed post-processing Identify best method and implement 45 K
Improve weekend data (LDV) Create weekend trip tables, validate/calibrate relative distributions 75 K
Link-level EFs Trucks from Lit or E55/E98 data, MOBILE6 for LDVs 50 - 75 K
Very High
High
Moderate
Low
Note cost assumptions in speaking notes window
20
Phase II priority projects

• 2010, 2015, 2020 on-road forecasts.
• BURDEN 2007 control totals
• Statewide model (rather than ITN) w/DTIM for
  spatial allocation
• Interpolate trip tables for intermediate year
  assignments.
• Disperse (via spline interpolation) the on-road
  allocation to approximate the impact of network
  elements not explicitly modeled in the Statewide
  network

21
Impact of Spline Function
Source Atm. Env. V.38, issue 2, 305-319 (2004)
22
Phase II priority projects

• Improve truck activity estimates
• Reverse fit an OD table to observed truck counts,
  use SJV Phase II goods movement model as an
  initial condition
• Base projections on TAZ employment growth
• Rational Heavy-duty truck activity is poorly
  understood.

23
Count Locations from Phase II truck Model
24
Phase II priority projects

• Speed post processing link data
• Post process speed data to represent hourly
  conditions
• Research into the sensitivity / appropriateness
  of different formulations
• SAS code to implement
• Impacts highly congested links
• Rational As shown in the literature review, the
  impact of speed post processing on estimated
  emissions can be dramatic for links operating
  near and over capacity conditions.

25
Phase II priority projects

• Improve weekend spatial allocation
• Incorporating behavioral characteristics into the
  method (e.g., ratio OD tables by trip type and
  ITE data).
• Reverse fit OD tables to observed light duty
  counts
• Rational Trip making patterns change along with
  trip generation rates for weekend activity.
  Currently only trip rates are taken into account

26
Phase II priority projects

• Link level emission rates
• Use emission rates and activity data with
  similar spatial specificity
• HDV emission rates from models in the literature
• Option use E55/E59 data to construct new rates
  based on Kear Niemeier 2006
• Use light duty rates from MOBILE6
• BURDEN 2007 still sets control totals
• Rational Link-based emissions rates are based on
  road segment level activity. BURDEN trip based
  rates include operation over all facility types

27
Q A

• What effect will time and resource constraints
  have on CCOS priorities?
• How does the on-road inventory uncertainty
  compare to that in the rest of the inventory?
• Different projects have different uncertainties
  and commensurate impacts
• Extrapolations from an inappropriate set of year
  2000 assumptions would have little value.
  Internal consistency and a scientific/behavioral
  bases for on-road activity is critical.