This document compares source apportionment approaches for PM10 using tagged species (TS) contributions and brute force impacts (BF) simulations with the CAMx and FARM air quality models. Non-linear responses of PM10 concentrations to emission reductions, particularly for agriculture, are highlighted by the dependence on emission reduction levels and differences between TS and BF results. The gas ratio and Stein and Alpert decomposition indicate relationships between changes in secondary inorganic aerosol formation chemistry and non-linear responses to precursor emission reductions. Differences between linear and non-linear sources suggest limitations in using TS to infer emission change responses.

1. JRC
Comparison of source apportionment approaches
8a Giornata della modellistica in ARIA(NET)
Belis C.A.1, G. Pirovano2, M.G. Villani3, G. Calori 4, N. Pepe4, J. P. Putaud1
1 European Commission, Joint Research Centre, via Fermi 2749, 21027 Ispra (VA), Italy
2 RSE Spa, via Rubattino 54, 20134, Milan, Italy
3 ENEA Laboratory of Atmospheric Pollution, via Fermi 2749, 21027 Ispra (VA), Italy
4 ARIANET s.r.l. via Gilino, 9 - 20128 Milan (MI) – Italy

2. Overview
In this study we
• compare PM10 tagged species (TS) contributions with brute force impacts (BF,
or emission reduction impacts) at different emission reduction levels (ERLs)
• analyse the geographical patterns
• compute interaction terms (of the Stein and Alpert algebraic expression) for the
studied sources
• examine the behaviour of PM10 in various areas (urban, rural, etc.) where
different chemical regimes prevail
The focus is on the SIA formation, with particular reference to ammonium nitrate
(NH4NO3) and ammonium sulfate ((NH4)2SO4).

3. Simulation with tagged species and brute force approaches
• Models: CAMx and FARM
• Tagged species module: PSAT
• Pollutant: PM10
• Area: Po Valley
• Reference year: 2010
• Time window: full year
• Domain: 580 x 400 km2
• Grid step: 5 km x 5 km approx.
• Meteorology: WRF 14 layers
• Emissions: EMEP (Europe), ISPRA (Italy), INEMAR (regional) processed with SMOKE

4. Simulations with Brute Force approach
Simulation set
Reduced sources
CAMx 100% CAMx 50% CAMx 20% FARM 50% FARM20%
No reduction Base case CAMx Base case FARM
AGRICULTURE (AGR) x x x x x
INDUSTRY (IND) x x x x x
ROAD TRANSPORT (TRA) x x x x x
RESIDENTIAL (RES) x x
AGR-IND x x x x x
AGR-TRA x x x x
IND-TRA x x x x x
RES-IND x x
RES-TRA x
RES-AGR x
AGR-IND-TRA x x x x
RES-IND-TRA x x
Simulations in blue are not discussed in this presentation

5. Annual contributions of PM10 sources, TS (CAMx PSAT)

6. Annual contributions of PM10 sources, BF
AGRICULTURE
ROAD TRANSPORT
INDUSTRY
FARM 50% x2 FARM 20% x5
CAMx 50% x2 CAMx 20% x5
CAMx 100%
6 10 14 6 10 14 6 10 14 6 10 14 6 10 14
longitude
47
46
45
44
latitude
47
46
45
44
47
46
45
44
% of base case
0 10 20 30 40 50

7. Annual contributions of PM10 sources, TS vs BF

8. Non-linear responses of PM10 concentrations to emission reductions (mainly
agriculture) highlighted by:
• Dependence on the emission reduction level (BF)
• BF impacts not proportional to TS contributions
Non-linear responses of PM10 concentrations to emission reductions (mainly
agriculture) investigated, based on:
• Stein and Alpert (1993) decomposition
• The “free ammonia” / “total nitrate” ratio (GR), Ansari and Pandis (1998)
Non linearity

9. Stein and Alpert decomposition (theoretical example)
The interaction terms ( መ
𝐶) of the factor decomposition measure the
consistency between the impacts obtained with single source
reductions compared to those of multiple source reductions and
therefore can be used as indicators of non-linearity.
Clappier et al., 2017
Stein and Alpert decomposition
2 sources
∆𝐶𝑨𝑩= ∆𝐶𝑨 + ∆𝐶𝑩 + መ
𝐶𝑨𝑩
Stein and Alpert decomposition
3 sources
∆𝐶𝑨𝑩𝑪= ∆𝐶𝑨 + ∆𝐶𝑩 + ∆𝐶𝑪 + መ
𝐶𝑨𝑩 + መ
𝐶𝐴𝐶 + መ
𝐶𝑩𝑪 + መ
𝐶𝐴𝐵𝐶
Stein and Alpert, 1993

10. The gas ratio (GR, Ansari and Pandis,1998) was used to define the chemical regime in
each of the simulations :
GR= ([NH3] + [NH4
+] – 2[SO4
2-]) / ([HNO3] + [NO3
-])
where concentrations are nmol.m-³ or in nmol.mol-1 of air (ppb).
The GR value defines three different chemical regimes:
(a) GR>1, in which NH4NO3 formation is limited by the availability of HNO3,
(b) 0<GR<1, in which NH4NO3 formation is limited by the availability of NH3, and
(c) GR<0, in which NH4NO3 formation is inhibited by H2SO4the gas ratio (GR)
“free ammonia” / “total nitrate” gas ratio

11. GR vs interaction terms in a rural site (Cremona province)
negligible interaction terms
interaction terms > 0
interaction terms < 0
C: CAMx and F: FARM.
10, 5 and 2 indicate the 100%, 50% and 20% ERLs, respectively.
A: agriculture, I: industry and T: transport
Legend

12. GR vs interaction terms in an urban site (Milan)
negligible interaction terms
interaction terms > 0
interaction terms < 0
C: CAMx and F: FARM.
10, 5 and 2 indicate the 100%, 50% and 20% ERLs, respectively.
A: agriculture, I: industry and T: transport
Legend

13. • The differences in PM10 attributed to AGR applying the TS and the BF approaches at 100%
ERL reach a factor 2.
• There is less but still considerable dispersion around the identity line between BF impacts and
TS contributions of AGR at 50% and 20% ERLs (indicating spatial inconsistencies).
• The association between AGR with non-linear responses is due to the key role of NH3 (mainly
emitted by this source) in the formation of secondary inorganic aerosol.
• TS and BF lead to comparable PM10 apportionments for the other studied sources IND, TRA
(and RES, not shown).
• The higher comparability between CAMx and PSAT is because are performed with the same
model, while FARM and PSAT result from two different models.
Conclusions 1

14. • The analysis of the interaction terms vs free ammonia / total nitrate ratios (GR) provides
evidence about the relationships between changes in the SIA formation chemical regime and
the non-linear response of PM10 concentrations to SIA precursor emissions reductions.
• As linearity is more important for daily than yearly averages, the differences will be even more
important for daily or hourly values.
• FAIRMODE recommendation: use of TS to infer the response to emission changes should be
limited to linear compounds
Conclusions 2

15. Thank you
© European Union 2020
Unless otherwise noted the reuse of this presentation is authorised under the CC BY 4.0 license. For any use or reproduction of elements that are not
owned by the EU, permission may need to be sought directly from the respective right holders.