Table 4 Overview of the estimated regression model specifications

From: Reduced ultrafine particle levels in São Paulo’s atmosphere during shifts from gasoline to ethanol use

Estimates reported in

Dependent variable(s) (and data source)

Sample descriptiona

Time aggregation

Fuel mix variable

Estimation procedure(s)

Other sensitivity analysis provided

Supplementary Table 1

PM2.5 mass concentration, 24-h filter (CETESB)

11/2008–5/2013, 3 sites every 6 days

24-h mean

Gasoline share in the flex fleet

OLS + bootstrap

Across the columns, more and alternative controls are introduced, e.g., trend and meteorology

Supplementary Table 2 (column 1 in Supplementary Fig. 10)

PM2.5 mass concentration, 24-h filter (CETESB)

11/2008–5/2013, 3 sites every 6 days

24-h mean

Gasoline share in the flex fleet, or aggregate fleet

OLS + bootstrap; 2SLS; OLS

Across the columns, the gasoline share and the estimation procedure are varied. Models include controls

→Tables 2 and 3 and Fig. 1 report the effect of raising the gasoline share on 24-h PM2.5 levels (model with wind direction controls). Figure 3 reports robustness to diesel control

Supplementary Table 3 (column 2 in Supplementary Fig. 10)

PM2.5 mass concentration, beta continuous (CETESB)

1/2011–5/2013, 3 sites by 2012

5-h mean, 07:00 to 11:00

Gasoline share in the flex fleet, or aggregate fleet

OLS + bootstrap; 2SLS; OLS

Across the columns, the gasoline share and the estimation procedure are varied. Models include controls

Supplementary Table 4 (column 3 in Supplementary Fig. 10)

Ozone mass concentration (CETESB)

11/2008–5/2013, 12 sites

5-h mean, 12:00 to 16:00

Gasoline share in the flex fleet, or aggregate fleet

OLS + bootstrap; 2SLS; OLS

Across the columns, the gasoline share and the estimation procedure are varied. Models include controls. Figure 3 reports robustness to diesel control

→Table 2 and Fig. 1 report the effect of raising the gasoline share on ozone levels in the early afternoon (reproduced in column 1 of Supplementary Table 4)

Supplementary Table 5

Particle count, nucleation, Aitken, accumulation, BC mass concentration (Own)

10/2010–5/2011 for DMPS, 1 site (similar periods for other param.)

24-h mean

Gasoline share in the flex fleet

OLS + bootstrap; 2SLS

Across the columns, the dependent variable and the estimation procedure are varied. Models include controls

→Table 3 reports the effect of raising the gasoline share on 24-h nucleation, Aitken, accumulation, BC levels (model with wind direction controls)

Supplementary Table 6 (Supplementary Fig. 13 shows panel D)

UFP 7‐100 nm (Own)

10/2010–5/2011, i.e., full period of field campaign, 1 site

1-h meanb

Gasoline share in the flex fleet

OLS + bootstrap; 2SLS

The panels show variation in the sample (all days of the week vs. weekdays only), the estimation procedure, and the effect of wind direction controls

Supplementary Table 7 (Supplementary Fig. 13 shows panel D)

PM 100‐800 nm (Own)

10/2010–5/2011, i.e., full period of field campaign, 1 site

1-h meanb

Gasoline share in the flex fleet

OLS + bootstrap; 2SLS

The panels show variation in the sample (all days of the week vs. weekdays only), the estimation procedure, and the effect of wind direction controls

Supplementary Table 8 (Supplementary Fig. 14 shows panels B,D)

UFP 7‐100 nm and PM 100‐800 nm (Own)

1/2011–5/2011, i.e., more seasonally homogeneous sample

1-h meanb

Gasoline share in the flex fleet

OLS + bootstrap

The panels show variation to including a linear trend vs. not allowing a trend (Specifications otherwise follow panel D, Supplementary Tables 6 & 7.c)

→Table 2 and Figs. 1 & 2 report the effect of raising the gasoline share on UFP 7–100 nm and PM 100–800 nm levels at 08:00 (reproduced in panels B and D of Supplementary Table 8).

→Figure 3 reports robustness to diesel control. Figure 4 shows effects over the day and across the size range using the specification in panels B and D of Supplementary Table 8.

Supplementary Figs. 11 & 12

Particle count and BC mass concentration (Own)

11/2010–5/2011 for CPC, 1 site (similar period for BC)

1-h meanb

Gasoline share in the flex fleet

OLS + bootstrap

 

→Table 2 and Fig. 1 report the effect of raising the gasoline share on BC levels at 08:00. Figure 3 reports robustness to diesel control.

Supplementary Fig. 9

PM2.5 mass concentration, beta continuous (CETESB)

1/2011–5/2013, 3 sites by 2012

1-h meanb

Gasoline share in the flex fleet

OLS + bootstrap

 
  1. aAll estimated samples exclude the colder months of June to September
  2. bHour-by-hour regressions
  3. cSample restricted to non-holiday weekdays, wind direction controlled for
Back to article page