In the format provided by the authors and unedited. SUPPLEMENTARY INFORMATION DOI: 10.1038/NGEO2960 Top-of-atmosphere radiative forcing affected by brown carbon in the upper troposphere Yuzhong Zhang 1, Haviland Forrister 1, Jiumeng Liu 2, Jack Dibb 3, Bruce Anderson 4, Joshua P. Schwarz 5, Anne E. Perring 5,6, Jose L. Jimenez 6,7, Pedro Campuzano-Jost 6,7, Yuhang Wang 1, Athanasios Nenes 1,8,9,10, Rodney J. Weber 1 * 1 School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia, USA, 2 Atmospheric Sciences and Global Change Division, Pacific Northwest National Laboratory, Richland, Washington, USA, 3 Earth Systems Research Center, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, New Hampshire, USA, 4 Chemistry and Dynamics Branch, NASA Langley Research Center, Hampton, Virginia, USA, 5 Chemical Sciences Division, Earth System Research Laboratory, National Oceanic and Atmospheric Administration, Boulder, Colorado, USA, 6 Cooperative Institute for Research in Environmental Sciences, University of Colorado Boulder, Boulder, Colorado, USA, 7 Department of Chemistry and Biogeochemistry, University of Colorado Boulder, Boulder, Colorado, USA, 8 School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA, 9 Institute of Environmental Research & Sustainable Development, National Observatory of Athens, Greece, 10 Institute for Chemical Engineering Science, Foundation for Research and Technology Hellas, Patra, Greece, *e-mail: rweber@eas.gatech.edu NATURE GEOSCIENCE www.nature.com/naturegeoscience 1 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Table S1. Date, inflow times, inflow altitude, outflow times, outflow altitude, and outflow temperature for all convective storms during the DC3 field campaign. All data except the outflow temperature are adopted from Ref. 25. Flight Date Inflow Times (UTC) Inflow Altitude (km) Outflow Times (UTC) Outflow Altitude (km) Outflow Temperature (K) 1 May 18 22:10 22:52 3.93 23:17 23:48 11.18 218 2 June 2 20:08 21:30 2.08 21:58 23:09 11.23 223 3 June 5 23:08 23:42 3.28 0:03 0:38 9.50 238 4 June 6 20:35 21:23 2.34 21:47 21:57 10.80 225 5 June 6 22:12 23:00 3.13 23:27 23:52 11.48 222 6 June 15 21:18 21:38 1.84 22:13 22:36 10.98 224 7 June 22* 22:34 23:55 2.80 0:16 1:39 10.81 229 8 May 19 22:35 0:14 3.06 0:40 1:00 10.39 227 9 May 25 23:42 0:54 3.04 1:17 1:28 11.33 225 10 May 29** 22:09 23:15 2.97 23:40 0:13 10.48 231 11 June 1 0:32 1:22 2.81 1:38 2:00 11.17 223 12 June 16 0:02 1:21 1.22 1:41 2:02 11.84 222 * The June 22 event is reported to have ingested a biomass-burning plume 23. ** The convective outflow from the May 29 event was also sampled downwind on May 30 23.
Table S2. Same as Table 1, except results are computed with the global-mean cloud distribution. Unit: W m -2 Net RF Absorption BrC+BC BrC BC Scattering Column -8.8±3.2 5.5±2.3 1.4±0.6 4.1±2.1-14.3±3.3 High altitudes (> 5 km) -1.9±1.9 2.6±1.4 0.91±0.47 1.7±1.2-4.5±2.1 Low altitudes (< 5 km) -6.9±2.3 2.9±1.2 0.47±0.22 2.4±1.2-9.8±2.2
Table S3. Same as Table 1, except results are computed with SEAC 4 RS data. Unit: W m -2 Net RF Absorption BrC+BC BrC BC Scattering Column -21.6±11.1 4.1±3.7 1.1±1.3 3.0±2.7-25.8±13.4 High altitudes (> 5 km) -5.8±7.1 1.2±1.9 0.51±0.70 0.7±1.3-7.0±7.8 Low altitudes (< 5 km) -15.8±8.2 3.0±2.5 0.59±0.80 2.4±1.9-18.8±10.2
Table S4. Measured variables and observation-constrained parameters used to compute the input for the radiative transfer model. Measured variables b SE,BrC (365nm) Description Method Eq. # BrC solution absorption coefficient at 365 nm. Spectrophotometric measurements of solution extracts from filter samples (3) M BC BC mass concentrations SP2 (4) b Scat (450nm) b Scat (550nm) b Scat (700nm) Observation-constrained parameters Aerosol scattering coefficients at 450, 550, and 700 nm TSI-3563 Nephelometer (5) f BrC Factor converting solution BrC absorption to aerosol BrC absorption Mie calculation with validation by multiple field studies (3) AAE BrC Spectral dependence of BrC absorption Linear regression fit of extract solution log(abs) versus log(λ) in wavelength range 300 450 nm (3) AAE BC Spectral dependence of BC absorption Literature (4) MAC BC (660nm) Mass absorption cross section of BC Linear regression fit of PSAP total aerosol absorption coefficient at 660 nm and M BC during DC3 (4)
Fig. S1. Vertical profile (medians and inter-quartile) of b BrC /b BC at 365 nm in DC3 (a and b) and SEAC 4 RS (c and d) over multiple regions in the U.S. a and c plot the locations and altitude (color) of the samples. Circles represent biomass burning samples and triangles non-biomass burning samples. b and d show the vertical profiles of b BrC /b BC at 365 nm by region.
Fig. S2. Dry matter burnt (in 10 6 kg) during (a) May June 2012 (period of DC3 campaign) and (b) August-September 2013 (period of SEAC 4 RS campaign). The data are from GFED 4s fire emission inventory and the horizontal resolution is 0.25 0.25.
Fig. S3. Same as Fig. 2, except results are computed with the global-mean cloud distribution.
Fig. S4. Absorption aerosol optical depth and aerosol radiative forcing in SEAC 4 RS and comparisons of upper tropospheric chemical environment between DC3 and SEAC 4 RS. (a) Same as Fig. 1C, except results are for SEAC 4 RS. (b) Same as Fig. 2, except results are computed with the SEAC 4 RS data. (c and d) Comparisons of median vertical profiles of CO (c) and NO x (d) between DC3 (red) and SEAC 4 RS (blue). Error bars represent interquartile ranges. To compare the regional background, we exclude samples identified as biomass burning plumes or convective storms.
Fig. S5. Among the 12 convection events studied, the June 22 event, which ingested a wildfire plume, features highest BrC absorption in the outflow. The median of outflow BrC absorption coefficients at 365 nm is plotted, in comparison with the median and interquartile ranges (error bar) of other events.
Fig. S6. Mean and standard deviation of b BrC at 365 nm in fresh convective outflow sampled on May 29 and its aged plume sampled on May 30.
Fig. S7. The radiative and climate impact of BrC may not be restricted to just the fire vicinity or other BrC emission sources. Convection processed (or produced) BrC may radiatively affect a large region.
Text S1. The direct RF computed under cloudy conditions Previous studies have shown that the direct RF of absorbing aerosols is also sensitive to the vertical distribution of clouds. Absorbing aerosols above clouds exert positive RF while those below clouds exert negative RF 43. Using a global-mean cloud distribution 43, we estimated the instantaneous direct RF of absorbing aerosols under cloudy conditions (Table S2 and Fig. S3). Compared with the clear-sky scenario, the RF due to absorbing aerosols (BC+BrC) increased by ~ 30% under cloudy conditions, half of which is contributed by high altitude absorbing aerosols, whereas the RF due to scattering is greatly reduced. As a result, the net cooling of the aerosols under cloudy conditions is only about half the cooling for clear skies. In particular, the net RF by aerosols is only slightly negative at high altitudes, demonstrating the increased importance of high altitude BrC under cloudy conditions.
Text S2. Comparisons of aerosol radiative impacts between DC3 and SEAC 4 RS Fig. S4B and Table S3 show the results of clear-sky instantaneous radiative forcing calculations constrained with SEAC 4 RS observations. The SEAC 4 RS campaign was undertaken over the continental U.S. in the summer following the DC3 study and used the same NASA aircraft with largely the same instrument payload. Similar to DC3 (Table 1 and Fig. 2), the RF due to BrC (1.1±1.3 W m -2 ) is about 25% of the RF due to BrC and BC combined (4.1±3.7 W m -2 ) in SEAC 4 RS. The contribution of high-altitude BrC to the BrC RF in the column is also significant in SEAC 4 RS (45%) but less pronounced than in DC3 (70%), which results from less high-altitude BrC observed in SEAC 4 RS (Fig. 1C and Fig. S4). Consistent with convection as a major source for highaltitude BrC, this difference between the two campaigns can be explained by less convection influence in the upper troposphere during SEAC 4 RS, which is supported by less upper tropospheric CO (from convective transport) (Fig. S4C) and NO x (from lightning) (Fig. S4D) observed in SEAC 4 RS. It is noteworthy that biomass burning emissions are not significantly different between the two campaigns over the sampling regions (Fig. S2), and thus are unlikely to explain the difference in high-altitude BrC levels.