1. Impact of New North American Emissions Inventories on Urban Mobile Source Emissions for High-Resolution Air Quality Modelling
Junhua Zhang, Qiong Zheng, and Michael Moran
Air Quality Research Division, ECCC, Toronto, Ontario, Canada
8th International Workshop on Air Quality Forecasting Research, Toronto, Ontario, Canada Jan. 2017

2. Outline Motivation for this study
Recent changes to on-road emissions processing
- base year
- emissions estimation tools
- spatial surrogates
- temporal profiles
- chemical speciation
Example of impacts of changes to processed on-road emissions for 2 high-resolution sub-grids covering 2 large North America metropolitan areas: New York city (NYC) and Greater Toronto and Hamilton Area (GTHA)
Summary and conclusions

3. Motivation
On-road mobile sources are an important source of emissions, especially in cities. In 2011, 23% of total anthropogenic NOx emissions in Canada were from on-road mobile sources and 40% in the US.
In recent years the tool to estimate on-road mobile emissions has changed from MOBILE6 to MOVES (MOtor Vehicle Emission Simulator)
MOVES has also undergone frequent updates from version to version 2014a, resulting in relatively large changes of on-road emissions estimates from inventory to inventory
Spatial and temporal allocation of on-road mobile emissions become increasingly important when AQ model resolution approaches the kilometer level
It is important to understand the impact of the above changes for high-resolution AQ modelling

4. Changes to Recent Canadian and US On-road Emissions Inventories (1)
Replacement of MOBILE6 by MOVES started with the preparation of the US 2005 (and projected 2012) and Canadian 2010 inventories
Unlike MOBILE, MOVES differentiates between on-roadway emission processes and off-network emission processes for engine starts, idling, fuel vapor venting, etc. In addition to 12 road types, a new road type, off-network, with large emissions was introduced in MOVES
MOVES-Based 2010 Canadian Monthly On-road NOx Emissions by Road Type

5. Canada On-road Inventory
Changes to Recent Canadian and US On-road Emissions Inventories (2)
NY State population-2015
19.8m
Ontario population-2015
13.9m
A new set of on-road Source Classification Codes (SCC) was introduced in MOVES2014 and SMOKE v3.6 (
Structure of Old SCC
Structure of New SCC
List of recently used/analyzed US and Canadian on-road inventories
US On-road Inventory
Canada On-road Inventory
Projected 2012v1
Gasoline: MOVES (Off-network emissions)
Diesel: MOBILE6
2010v1
Light-duty: MOBILE6.2C
Heavy-duty: MOVES2010b
Projected 2012v2
MOVES2010
Off-network emissions
2010v2
MOVES2010b
2011v1
2013v1
MOVES2014
New on-road SCCs
2011v2
2013v2
Projected 2017
2013v3
Inventories highlighted in red in the table and filled with zigzag lines in the plot are those that have been used for the Canadian operational AQ prediction system

6. Surrogate Description
Changes to Spatial Surrogates
– surrogates for old SCCs
12 road types (6 rural + 6 urban  6 surrogates (3 rural + 3 urban)
Road Type
Surrogate
Surrogate Description
Rural Interstate
210
Rural Primary Road Miles
Rural Principal Arterial
Rural Minor Arterial
Rural Major Collector
230
Rural Secondary Road Miles
Rural Minor Collector
Rural Local
130
Rural Population
Urban Interstate
200
Urban Primary Road Miles
Urban Freeway
Urban Principal Arterial
Urban Minor Arterial
Urban Collector
220
Urban Secondary Road Miles
Urban Local
120
Urban Population

7. 4 surrogates for 4 types of roads 11 surrogates for off-network SCCs
Changes to Spatial Surrogates
– surrogates for new SCCs (MOVES2014)
5 road types (2 rural + 2 urban + 1 off-network)
4 surrogates for on-network (2 rural + 2 urban); 11 surrogates for off-network for 2011V2 and projected 2017
various surrogates such as population also used previously for off-network emissions
4 surrogates for 4 types of roads
Road type 02 (rural restricted)
210 (rural primary roads)
Road type 03 (rural unrestricted)
231 (rural unrestricted roads)
Road type 04 (urban restricted)
200 (urban primary roads)
Road type 05 (urban unrestricted)
221 (urban unrestricted roads)
11 surrogates for off-network SCCs
Refueling
600 (gas stations)
EXT/APU (Hoteling)
205 (truck stops)
Source types 11/21/31 (passenger cars/trucks, motorcycles)
535 (res + comm + ind + inst + govt)
Source type 32 (light comm. trucks)
510 (commercial + industrial)
Source type 41 (intercity buses)
258 (intercity bus terminals)
Source type 42 (transit buses)
259 (transit bus terminals)
Source type 43 (school buses)
506 (education)
Source type 51 (refuse trucks)
875 (landfills)
Source types 52/61 (short haul trucks)
256 (off-network short haul trucks)
Source types 53/62 (long haul trucks)
257 (off-network long haul trucks)
Source type 54 (motor homes)
526 (residential non-institutional)

8. Changes to Temporal Profiles - temporal profiles for old SCCs
E.g., 12 weekday and 12 weekend diurnal profiles for 12 road types

9. Changes to Temporal Profiles - temporal profiles for new SCCs
Very detailed weekly and hourly profiles were created based on U.S. VTRIS (Vehicle Travel Information System, national/state/county statistics.
For NY state there are 52 weekly profiles and 156 diurnal profiles. In addition to weekday and weekend profiles, day-specific and vehicle-specific diurnal profiles have also been created and are used

10. Changes to Chemical Speciation
Many new on-road VOC speciation profiles have been introduced in recent SPECIATE databases. For example, SPECIATE 4.4 (released Feb. 2014) has 43 new exhaust/evaporative gas profiles for on-road mobile emissions and SPECIATE 4.5 (released Sept. 2016) has another 33 exhaust-gas profiles
Previously only one NOx speciation profile was used (NO:0.891, NO2:0.1, and HONO:0.009)
Recent US inventories (2011v2, projected 2017) based on MOVES2014 include pre-speciated SCC-specific NOx emissions: NO, NO2, and HONO, in addition to NOx emissions (beware potential for double-counting)
Not all of the new VOC speciation profiles have been used for emissions processing, but they are available to be used

11. Model Grid and Analysis Subgrids
Consider two 60x60 subgrids of PanAm km high-resolution AQ modelling grid
Subgrids cover the metropolitan areas of Toronto (GTHA) and New York city (NYC)
GTHA subgrid,Population 8M
NYC subgrid, Population 19M

12. Seven Emissions Cases Considered
(4 for the NYC subgrid, 3 for the GTHA subgrid)
Case
Description
NYC-C1
Projected 2012v1 US inventory, 6 spatial surrogates for 12 road classes, off-network emissions, for gasoline vehicles, old temporal profiles: used previously for Canadian Regional Air Quality Deterministic Prediction System (RAQDPS)
NYC-C2
2011v1 US inventory, 6 spatial surrogates for 12 road classes, population surrogate for off-network emissions, old temporal profiles: now used for RAQDPS
NYC-C3
2011v2 US inventory (based on MOVES2014), 4 spatial surrogates for 4 road classes, 11 spatial surrogates for off-network SCCs, new temporal profiles
NYC-C4
Same as NYC-C3 except based on projected 2017 US inventory: being tested as potential candidate for RAQDPS
GTHA-C1
2010v1 Cdn inventory, 6 spatial surrogates for 12 road classes, no off-network emissions, population used for 2 local road surrogates, old temporal profiles
GTHA-C2
Same as GTHA-C1 except capped population (Gately et al., 2013, ES&T, 47, ; Moran et al., 2015, 14th CMAS Conf.) used for 2 local road surrogates: now used for RAQDPS
GTHA-C3
2013v3 Cdn inventory, 4 road surrogates for 4 road classes, capped population surrogate used for off-network emissions, old temporal profiles: being tested for potential candidate for RAQDPS
Inventories in red have been used for RAQDPS and the two in red are being tested for potential candidates for RAQDPS.
Traffic emissions are inversely related to population density above 2000 person km-2 (Gately et al., 2013, ES&T, 47, ) based on statistics over 351 Massachusetts towns done by Conor Gately from Boston University (BU), who is doing CO2 emissions inventory.

13. Subgrid Total NO Emissions for July for Seven Cases
NYC subgrid population 19M
GTHA subgrid population 8M
Heather Simon, 2016:
To improve model agreement, some researchers have proposed reducing mobile-source NOx by 30% to 70%
Variation of subgrid total NO emissions is consistent with variation of input emissions inventories shown on Slide 5
Difference of the subgrid total NO emissions between the two subgrids also reflects the difference in total population between them
NOx emissions are reduced by 40% from the 2011v2 inventory to the projected 2017 inventory for the NYC subgrid. This will help to reduce the reported high model NOx (NOy) bias (e.g., Simon, et al., 2016, 15th CMAS Conf.) when the 2011 US inventory is used

14. Current EPA 2011 CMAQ/CAMx NOx and NOy Evaluation (slide from Simon et al., 2016)
CMAQv5.0.2
NOx bias – all AQS sites
CMAQv5.1
NOx bias – all AQS sites
CAMx v6.2
NOx bias – all AQS sites
NOx is generally unbiased or under-predicted during daytime but is over-predicted in morning and evening transition hours and at night
NOy tends to be over-predicted by the models at all times of day
CMAQv5.1 has improved characterization of mixing in morning/evening transitions and at night compared to CMAQv5.0.2
NOx and NOy biases decrease in CMAQv5.1 versus CMAQv5.0.2
CAMx v6.2 biases in NOx and NOy generally fall between biases from CMAQv5.0.2 and v5.1
CMAQv5.0.2
NOy bias – all AQS sites
CMAQv5.1
NOy bias – all AQS sites
CAMx v6.2
NOy bias – all AQS sites
This slide was from Heather Simon’s presentation (2016 CMAS) to show that NOx/NOy was overestimated by CMAQ and CAMx using 2011 NEI

15. Spatial Distribution of July NO Emissions - NYC Subgrid (1)
Case NYC-C1:
Projected 2012v1 US inventory (released in )
Emissions were mainly distributed along major highways

16. Spatial Distribution of July NO Emissions - NYC Subgrid (2)
NYC-C1: Projected 2012v1
NYC-C2: 2011v1
For NYC-C2 vs. NYC-C1, more emissions allocated to populated areas due to the large off-network emissions, for which population was used as spatial surrogate
For NYC-C3, 11 spatial surrogates were used for off-network emissions, resulting in less emissions than NYC-C2 but more emissions than NYC-C1 in populated areas
NYC-C4 has significantly less emissions than other 3 cases due to the reduction of NOx emissions in projected 2017 inventory
NYC-C3: 2011v2
NYC-C4: projected 2017

17. Spatial Distribution of July NO Emissions - GTHA Subgrid (1)
Case GTHA-C1:
2010v1 Cdn inventory (released in 2013)
Similar to NYC-C1 case, emissions were mainly distributed along major highways

18. Spatial Distribution of July NO Emissions - GTHA Subgrid
GTHA-C1: 2011v1
GTHA-C2: 2011v1
For GTHA-C2 vs. GTHA- C1, use of capped population as surrogate for emissions from local roads reduced emissions by 20%-50% in densely populated areas
For GHTA-C3 more emissions were allocated to populated areas than for GTHA-C1 and GTHA-C2 due to the large off-network emissions, even though capped population was used as spatial surrogate for off-network emissions
GTHA-C3: 2013v3
GTHA-C2/GTHA-C1

19. Time Series of NO Emissions - NYC Subgrid
NYC Subgrid Total NO Emissions – One Week
Time series of the 2011v1 emissions is very similar to that of the projected 2012v1 emissions, although spatial distribution is much different
Time series of 2011v2 emissions is very different from the time series of 2011v1 due to different temporal profiles used
Time series of projected 2017 emissions is similar to 2011v2, but day-time emissions were reduced by about 50%
Weekday Hourly NO Emissions
Relative Difference between 2011v1 (C2) and v2 (C3) 0%

20. Time Series of NO Emissions - GTHA Subgrid
GTHA Subgrid Total NO Emissions – One Week
Diurnal variation of emissions is similar between the 2010v1 and the 2013v3 inventories, although spatial distribution is quite different
NOx emissions are increased by about 15% over the GTHA when the 2013v3 inventory is used

21. Canadian Diurnal Profiles were created from FEVER Study (Zhang et al
Canadian Diurnal Profiles were created from FEVER Study (Zhang et al., 20th EI conference, 2012), different from EPA profiles for HDV

22. Summary and Conclusions
Recent changes to Canadian and US on-road mobile emissions inventories have significant impacts on model-ready urban mobile emissions in terms of magnitude, spatial and temporal allocation, and chemical speciation
The projected 2017 US inventory released by the US EPA is likely to help reduce the large positive NOx bias predicted by air quality models using the 2011 US inventory
Due to changes in spatial surrogates, a larger portion of the mobile emissions will be allocated to populated areas
Changes to US temporal profiles will increase on-road emissions during the morning rush hours but reduce emissions during the afternoon rush hours and overnight
Whether these changes will improve AQ model predictions in North American urban areas still needs to be tested, particularly by high-resolution models

23. QUESTIONS?