Fleet and Activity Data with Improved Spatio-Temporal Details for Emissions and Air Quality Modeling

Focus Area

Air Quality


Air Quality






2-3 years

Research Idea Scope

Summary Objective: The objective of this research is to provide guidelines and best practices for practitioners for developing spatial and temporally detailed vehicle fleet and activity estimates for input to emissions and air quality models. The guidelines should provide methods for using widely-available data sources to develop gridded hourly emissions estimates suitable for input to air quality models. This problem statement was developed in response to the key research needs identified by the Regional Air Quality Subcommittee of the TRB ADC20 (Transportation-Air Quality) full committee, during the TRB 2015 meeting. Research conducted for NCHRP Project 25-38, Input Guidelines for the MOVES Model, resulted in guidelines for practitioners on how to develop inputs for MOVES for the purpose of developing regional and project-level inventories and/or emission factors. However, these guidelines do not address practices for developing inputs on a detailed (sub-county) spatio-temporal basis – i.e., by geographic unit (grid cell or TAZ) and time of day (hour). Spatio-temporal allocation of vehicle emissions may be desired for reasons including: • To create inputs to advanced air quality models. These models are run by the EPA, state air quality management agencies, regional planning offices, and environmental consulting companies to support the design and implementation of air regulations and to evaluate the attainment of air quality standards. • To characterize air pollution health impacts and population exposure, for public health protection and environmental justice assessment. Given the proximity of vehicle emissions and urban populations, spatial and temporal variability in emissions can pose direct health risks. The Sparse Matrix Operating Kernel Emissions (SMOKE) model is a software product developed by the U.S. EPA to spatially, temporally, and chemically allocate emissions from all natural and anthropogenic sectors. In the case of vehicle activity, SMOKE allocates MOVES emissions outputs to ensure that roadway characteristics are correctly assigned to model grids, and that the chemical speciation of emissions aligns with the chemical mechanism selected for the air quality simulation. SMOKE also includes a default temporal allocation, although for most traffic-related emissions, locally specific data should be used rather than SMOKE defaults. Dispersion models such as CAL3QHC and AERMOD have also been used for localized modeling of the air quality impacts of spatially disaggregated emissions. Spatio-temporal estimation of emissions requires consideration of start and evaporative emissions in addition to running emissions. The number of vehicle-starts by time period and geographic area must be known, as must the distribution of soak times of the started vehicles (time since being turned off). The distribution of VMT by road type and speed in the geographic area must also be known. Information about the vehicle fleet operating in the zone is also needed. Locations may have different emission rates per VMT or start depending upon the age and source type distribution of vehicles operating in the zone. A variety of data sources can be used to disaggregate vehicle emissions to a grid cell or TAZ level. State motor vehicle registration databases can provide vehicle population, age, and source type at a ZIP code level of detail, or even finer if geocoded locations are available in anonomized format. Metropolitan and statewide travel demand models can provide trip-ends and VMT by TAZ, often by up to four daily time periods. Activity-based travel models, only available in a few areas, can provide refined information on start and soak times. Household travel surveys can be mined to identify trip start and end times and locations at a finer level of temporal detail than represented in demand models. Cellphone and GPS data from private providers could potentially be used for a similar purpose, with much larger sample size and continuous sampling (but without the personal, household, and vehicle data contained in a travel survey). Dynamic traffic assignment models, where they are implemented, provide a “best practice” for considering the impacts of local traffic conditions on emissions. The most common approach in the past has been to apply emission factors to trip origins and VMT data by TAZ from the regional travel demand model. This was done using procedures to derive separate start- and VMT-based emission factors from the MOBILE model. The MOVES model produces start and running emissions in a different format, so many agencies may not be familiar with the best way to properly aggregate the running emissions rates that come out of MOVES and covert the grams/vehicle non-running emission rates to grams/start, grams/extended idle hour, etc. MOVES also aggregates evaporative emissions over the drive-cycle, making it difficult to assign them spatially. MOVES can produce various types of evaporative emissions for project-level analysis with specified soak time distributions, but agencies may not know how to merge this information with running emissions for regional-scale analysis. Accurate estimation of local emissions also requires information on the vehicle fleet specific to each area. This can require combining multiple data sources, including vehicle registration records, census data, and/or household travel survey data, to estimate local fleet characteristics. Many agencies may not be familiar with procedures for how to do this. To assist agencies with using the MOVES model for spatial and temporal disaggregation of emissions data, the following research activities are envisioned: 1. Review current practices for spatio-temporal disaggregation of emissions, and needs for improved methods for doing so. 2. Summarize and document best practices. 3. Develop methods that practitioners can apply with different levels of data and resource availability. 4. Document methods (including case studies/examples) in a reference document. 5. Provide software tools, such as sample scripts, that may be helpful in data processing.

Urgency and Payoff

Spatially disaggregated emissions inputs are required in areas that do regional air quality modeling analysis for State Implementation Plan and conformity purposes. Furthermore, the literature demonstrates a growing interest in evaluating the health impacts of transportation actions, including health effects of air quality changes. Air quality agencies probably need the results the most to inform regional air quality modeling. However, transportation agencies are probably best qualified to perform the work (especially the travel demand model part) and could also benefit from the spatial emissions and air quality information provided by the results, for example, for assessing environmental justice impacts. Also, regional air quality modeling can be an important driver of transportation emissions budgets required to achieve state implementation plans. This project could help support improved collaboration between transportation and air quality agencies for SIP and conformity analysis and compliance. (Note, this research statement was identified during the January 2016 annual TRB conference as one of the top priority regional air quality research needs. The statement was originally prepared by a committee that included: Chris Porter and David Kall (Cambridge Systematics, Inc.); Eulalie Gower-Lucas (Metropolitan Washington Council of Governments); Reza Farzaneh (Texas Transportation Institute); and Tracey Holloway (UW-Madison), on behalf of the TRB Air Quality Committee, Subcommittee on Regional Transportation and Air Quality Issues.)

Suggested By

Douglas Eisinger (on behalf of ADC-20 Regional Air Quality) Sonoma Technology, Inc. 707-665-9900

[email protected]