Overview #
The Utah Transit Authority (UTA) operates buses, light rail (TRAX), commuter rail (FrontRunner), and a streetcar across a six-county service area on Utah’s Wasatch Front — a rapidly urbanizing conurbation of approximately 1.8 million people with chronic air quality problems. Much of the service area is designated nonattainment or maintenance for PM₂.₅, ozone, CO, SO₂, and PM₁₀. Transit systems are widely assumed to benefit air quality (among other public goods), but the net effect depends on the balance between emissions generated by transit vehicles and emissions avoided when riders replace personal vehicle trips. Prior work lacked the spatial and temporal resolution to evaluate that balance at the scale of individual routes, hours, or neighborhoods.
This study modeled the full net emissions impact of UTA’s 2016 operations at hourly resolution and approximately city-block spatial scale (~200 m grid), covering vehicle miles traveled, fuel consumption, and five pollutant species — CO₂, CO, NOx, PM₂.₅, SOx, and NMHC. It was the first study of a complete US transit system to combine GTFS service schedule data with electronic fare collection (EFC) ridership records and automated passenger counter (APC) data to model both realized system emissions and gross avoided emissions from displaced personal vehicle trips simultaneously, at this resolution.
Region & Ecosystem #
The study area spans the full UTA service region along the Wasatch Front, from Ogden and Weber County in the north through Salt Lake Valley — the geographic and demographic core — to Utah County and Provo/Orem in the south. The Salt Lake Valley is a topographically enclosed airshed bounded by the Wasatch Mountains to the east, the Oquirrh Mountains to the west, and the Great Salt Lake to the northwest. Its topography drives severe wintertime temperature inversions that trap PM₂.₅ at ground level and produce some of the worst air quality episodes in the western United States. Summer ozone exceedances are a regular occurrence. These conditions give the question of transit’s net emissions impact direct public health relevance.
Methods #
The study estimated three quantities for each transit mode (bus, light rail, commuter rail): realized emissions from UTA vehicle operations, gross avoided emissions from personal vehicle trips displaced by transit ridership, and net emissions as the difference between the two — all resolved at hourly temporal and ~200 m spatial resolution.
GTFS spatial pipeline. UTA’s GTFS feeds specify the scheduled movement of every vehicle trip in time and space, but the raw data required substantial processing before it could support spatial emissions allocation. For each GTFS feed (UTA publishes updated feeds multiple times per year), unique route–direction–service combinations were identified from the trips table. Route line geometries were constructed from the shapes table in ArcGIS 10.3.1 and attributed with route, direction, and service identifiers. Stop locations were snapped to the route geometries they serve to enforce spatial coincidence. Routes were then segmented at stops, producing a set of stop-to-stop mini-trip segments. These segments were intersected with a regular 0.002° × 0.002° grid — the atmospheric modeling grid in use for concurrent University of Utah air quality research — generating a master table recording, for each segment, the route, direction, service, origin and destination stop IDs, and the grid cells traversed with distance through each cell. GTFS spatial data quality issues — including corner-cutting geometries, route extensions beyond terminal stops, and stops displaced from their routes — were diagnosed, catalogued, and reported to UTA.
Realized transit emissions. Vehicle-level emissions were estimated using UTA’s Transit Emissions Quantifier Tool, an adapted version of the APTA TEQ v10 parameterized with Utah fleet–specific emission factors. Bus services were disaggregated by four subregional business units (Central, Meadowbrook, Ogden, Timpanogos) with distinct fleet compositions. Light rail (TRAX) indirect emissions from off-site electricity generation were calculated but not allocated spatially, as Utah’s electricity is generated outside the Salt Lake Valley airshed.
Avoided emissions. EFC ridership records — tap-on and tap-off events for approximately 47–53% of boardings by mode — were joined to GTFS routes to reconstruct each rider’s path through space. The full EFC dataset was scaled to system-wide passenger miles using annual APC correction factors by mode. Each observed EFC trip was modeled as a displaced single-occupancy vehicle trip following the same route, with avoided emissions calculated using EPA average emission factors for the regional vehicle fleet. Emissions for all trip segments were binned to the hour of trip origination and allocated to the corresponding grid cells.
Fleet modernization scenario. A sensitivity analysis modeled the emissions impact of UTA’s planned fleet upgrades: replacement of older diesel buses with CNG and EPA clean diesel (2010+) vehicles, and upgrade of FrontRunner commuter rail locomotives from Tier 0+ to Tier 3.
Outputs #
- Annual, seasonal, monthly, and hourly net VMT, fuel consumption, and emissions estimates by mode and grid cell for 2016
- Spatial maps of net avoided travel and emissions for bus, commuter rail, and light rail (annual average hourly)
- Spatiotemporal PM₂.₅ emission maps by hour of day for current and projected future bus fleet
- Fleet modernization sensitivity analysis: projected emissions reductions under CNG/clean diesel bus and Tier 3 locomotive upgrades
- Published article: Environmental Research Communications, 2019; DOI: 10.1088/2515-7620/ab3ca7
My Role #
The analysis depended on a spatial data pipeline that could translate UTA’s GTFS feeds and EFC ridership records into geolocated, time-stamped emissions estimates at city-block resolution. I co-developed the study design with lead author Daniel Mendoza and conducted the literature review, drafting that section of the paper independently. I then designed and built the spatial data pipeline. Working in ArcGIS 10.3.1, I processed each GTFS feed to construct route line geometries, identify the full set of unique route–direction–service combinations, snap stop locations to their corresponding route geometries, and segment routes at stops to produce the stop-to-stop mini-trip segments that form the backbone of the emissions allocation. I then intersected those segments with the atmospheric modeling grid to generate the master lookup table the emissions calculations depended on. I processed the EFC ridership data against the same GTFS route representations to spatially allocate avoided passenger miles. I diagnosed and catalogued spatial quality issues in UTA’s GTFS data and presented findings and recommendations directly to UTA staff.
An undergraduate GIS intern worked on portions of this pipeline under my direct supervision in the iUTAH Visualization Laboratory, which I directed. All analytical design decisions, the overall pipeline architecture, and the work delivered to the lead author were my responsibility. I also assisted with data analysis, coauthored the manuscript, and presented results at conferences. Emissions factor selection, the avoided emissions modeling framework, and statistical analysis were led by the first author (Daniel Mendoza, Dept. of Atmospheric Sciences).