Black Carbon & Other Light- Absorbing Particles in Snow in Central North

Black carbon & other light- absorbing particles in snow in Central North America

& N. China

Sarah Doherty JISAO, Univ. of Washington Seattle, WA USA

Stephen Warren, Dean Hegg & Cheng Dang Dept. of Atmos. Science, Univ. of Washington, Seattle, WA USA

Issue: Black carbon in snow Why does BC in snow matter?

BC in snow lowers snow albedo (reflectivity) so more sunlight is absorbed by the snowpack snowpack warms snow grain size increases albedo lowered further snow warms further melts sooner concentrations of BC in surface snow increase further more albedo reduction accelerated snow melt

Net effects: • forcing (direct albedo reduction) & feedbacks (see above) lower surface albedo warms climate • earlier snowmelt impacts for agriculture, runoff timing

~2-30 ng/g BC ~2-30 ng/g BC ~1100ng/g BC ~2-30 ng/g BC ~1100ng/g BC

???? ng/g BC dust/soil? algae? Motivation Flanner et al., 2007 Fig. 5

• Focus has mostly been on BC in snow in the Arctic, BUT: • The highest concentrations of BC in snow are at lower latitudes • The open plains regions of the northern mid-latitudes are where the snowpack is not masked by vegetation • Warming due to BC in snow at lower latitudes may contribute significantly to Arctic warming (increased heat advection into Arctic) Motivation Qian et al., 2009 Regional model study of Western U.S. (Qian et al., 2009):

• decreases in snow accumulation rate

• increased runoff in February; decreased runoff March onward

• affects on mountain snowpack & snowpack in agricultural regions Motivation Motivation Large-area surveys of: Arctic (mostly 2007-2010) previous work under NSF N. China Great Plains (2010 & 2012) w/ Lanzhou Univ., China N. America Great Plains (2013 & 2014)

All using the same sampling & analysis technique Activities Measure BC and other insoluble light-absorbing particles in snow across the U.S. Great Plains Process samples from: - N. China survey (w/ colleagues from Lanzhou Univ.) - north-central Utah (w/ colleagues from PNNL) - Dye-2 Greenland: study effects of melt on surface BC Determine sources of light-absorbing particles in snow - U.S. Great Plains & China data sets

Test method of measuring BC vs. other light-absorbing particles - against another method of measuring BC (SP2) - by serial extraction of organics & iron analysis Study processes driving variations in surface snow mixing ratios

Measurement/model comparisons Activities Measure BC and other insoluble light-absorbing particles in snow across the U.S. Great Plains Process samples from: - N. China survey (w/ colleagues from Lanzhou Univ.) - north-central Utah (w/ colleagues from PNNL) - Dye-2 Greenland: study effects of melt on surface BC Determine sources of light-absorbing particles in snow - U.S. Great Plains & China data sets

Measurement/model comparisons N. American survey 2013 : 67 sites + 3 process study sites in 2014

2013

Site 1: 10 Jan

Sites 2-67: 28 Jan – 21 Mar

>500 snow samples

Canada 2014

Sites 68-70: 27 Jan – 24 Mar 2014 sites 2013 sites >360 snow samples U.S.

Vernal, Utah: 2013 & 2014 sampling by PMEL (J. Johnson & T. Quinn) MODIS Snow Cover (%) Feb 2013 FIELD SAMPLING

• ~2-5cm vertical resolution • 3 parallel profiles • collect soil at each site • melt/filter every ~3 days • re-freeze snow water for chemical analysis

• nuclepore filters 0.4μm pore size ~95% capture efficiency ISSW analysis of filters

...extrapolate down − Åabs

a τ ⎛ λ ⎞ to 300nm a,λ = , τ ⎜ ⎟ τ a,λo ⎝ λo ⎠

calculate Åabs (color) for λ=450, λ =600 o Åabs 1 black 2-4 brown ~6 reddish ISSW analysis of filters

Assume: Åabs=1.1 for BC Åabs site-specific for non-BC Partition absorption to get estimated BC via calibration curves (absorption mass)

est C BC ISSW analysis of filters

Scale by downwelling solar radiation ISSW analysis of filters

absorption of sunlight by non-BC constituents (organic carbon, soil, mineral dust)

f est absorption of nonBC sunlight by BC Derived Parameters:

est (ng/g) = estimated BC concentration C BC

equiv (ng/g) = amount of BC needed to account for all light CBC absorption 300-750nm (solar spectrum weighted) est (%) = fraction of 300-750nm solar absorption fnonBC due to non-BC components

[450:600nm] effectively the color Åtot (Åabs = 1 black 2-4 brown >~6 reddish)

Surface-most snow samples : BC mixing ratio est 250 ° 200 60 N C CBCest 150 BC (ng/g) 100 75 50

° 25 50 N 15 10

° 40 N

° W 120° 1 W ° ° W 90 110 W 100 Surface-most snow samples : Absorption Ångström exponent

3.6 brown ° 60 N 3.4 ÅÅ tottot 3.2 3 2.8 2.6 ° 50 N 2.4 2.2 2 1.8 1.6 ° 40 N 1.4 1.2 ° W 120° 1 black W ° ° W 90 110 W 100 Surface-most snow : non-BC light-absorbing particles expressed as an equivalent BC mixing ratio

250 equiv ° equiv C est 200 60 BC N C x f est 150 BC nonBC f × NBC (ng/g) 100 75 50

° 25 50 N 15 10

° 40 N

° W 120° 1 W ° ° W 90 110 W 100 site#: ï ï ï ï ï the mixingratioof BCinsnow

Pac Pac NW Vertical variabilityin mtn NW Intra- C BC est US Great Plains

Canada

250 200 150 100 75 50 25 15 10 5 1

W W °

90 est BC est BC C (ng/g)

W W

° 100

110

W ° N ° : 60 120 N °

50 N °

40 Surface-most snow samples : BC mixing ratio surface snow BC mixing ratios important for radiative forcing, albedo reduction

250 200 150 100 75 50 25 15 10 5 1

W W °

90 est BC est BC C (ng/g)

W W

° 100

110

: ° N °

60 120 N °

50 N °

40 *cold snowpacks only constraining important for Snow column average : BC mixing ratio BC mixing ratio * BC mixing ratio snowpack average snowpack average seasonal deposition * Positive Matrix PMF* Source “fingerprints” Factorization

#'%** '$$,+#'& '#$ +$% $+ )* ,% $+ )*

• Analyses done for one surface, one sub-surface sample at each site ( # *'& &+)+#'& ( # *'& &+)+#'& • Robust solutions not reached for all samples

ï ï %. ï* # $ # # & ) * $ -' , &#+)+ & *,$!+ '.$+ "$')# !')%+ *,ï%$ &'&ï& %&&#+'$ &'& &'& serial extraction ISSW analysis chemical analyses

for organics PMF Analysis : Factor contributions to 650-700nm absorption – Surface snow samples

Biomass ° Pollution 55 N Soil

° 50 N

Baaken Oil Fields ° 45 N

° 40 N

125° ° W W ° ° 120 ° W W 115 ° W ° ° W 90 110 W 105 W 100 95 PMF Analysis : Factor contributions to 650-700nm absorption – sub-surface snow samples

Biomass ° Pollution 55 N Soil

° 50 N

Baaken Oil Fields ° 45 N

° 40 N

125° ° W W ° ° 120 ° W W 115 ° W ° ° W 90 110 W 105 W 100 95 Combining N. American survey & earlier Arctic survey data 2009 Canadian Arctic survey

2007 Canadian sub-Arctic traverse

2013 N. Amer. Great Plains survey 60 est 40 C BC

20 BCest (ng/g)

0 2.5 Å 2.0 tot tot Å 1.5

1.0 50 55 60 65 70 75 80 latitude (oN) Re: the relative roles of soil vs. BC in US Great Plains snow albedo

Great Plains soil contribution is higher in sub-surface samples likely because this corresponds to when snowpack was shallower, so more exposed soil

Snow cover in 2013 was not anomalous – but there are years with more extensive & persistent snow cover. In these years, the relative role of BC (vs. soil) in lowering snow albedo will likely be higher

i.e., BC likely only dominates snow albedo reduction in years with higher snowpack – when retention of the snow is less critical for water resources Why so much soil in Sern Great Plains snow?

• Almost the entire area is (“snirt”)agricultural = disturbed soil • It’s windy (!!!) in the winter • Snow is often thin / patchy • Snow cover is intermittent, especially to the S and W Dirt mixes in with snow as it’s falling, right near the surface. Regional/global models will not capture this. Why so much soil in Sern Great Plains snow?

• Almost the entire area is (“snirt”)agricultural = disturbed soil • It’s windy (!!!) in the winter • Snow is often thin / patchy • Snow cover is intermittent, especially to the S and W Dirt mixes in with snow as it’s falling, right near the surface. un-tilled field Regional/global models will not capture this.

Farming practices may affect the color of snow at least as much as BC emissions in field tilled in the fall much of the southern Great Plains

Increased soil disturbance Bakken Oil fields • clearing for oil platforms • much more driving on dirt / farm roads • areas cleared for housing Increased BC emissions • diesel trucks • oil flaring (significant?) • wood stoves in temporary housing? N. China survey 2010 : 46 sites Lanzhou Univ. & Univ. of Washington collaboration (Wang et al., 2013)

Russia

Mongolia China

N Korea Beijing

S Korea Arctic + N. China + N. American surveys

Figure: Cheng, D.et al., J.Geophys. Res., (in preparation). N. China data: Wang et al., 2013 Arctic data: Doherty et al., 2010 N. Amer. data: Doherty et al., 2014 & 2016 2014 process study: 3 Idaho mountain valley sites

McCallMcCall CascadeCascade ValleyValley GardenGarden Valley Valley 80 1000 Cequiv 500 BC equiv 30 30 60 CBC 200 100 20 20 50 40 25 10 10 10 20 5

0 0 0 1 Jan Feb Mar Feb Mar Feb Mar 80 1000 Cest′ 500 BC est 30 30 60 C BC 200 100 20 20 50 40 25 10 10 10 20 5 Snow sample height (cm) Snow sample height (cm) 0 0 0 1 80 3.8 Å tot 30 30 3.4 60 Åtot 3 20 20 40 2.6 2.2 20 10 10 1.8 1.4 0 0 0 1 40 60 80 30 35 40 45 50 55 60 65 30 35 40 45 50 55 60 65 Day of year Day of year Day of year Overall findings

Dust & soil play a very strong role (sometimes dominate) incidences of high snow particulate light absorption at: US Great Plains sites 2 Idaho mountain valley sites SE of Vernal, Utah central China sites for the US GP & Idaho sites probably locally transported soil, so will not be captured by regional/global models radiative forcing by BC in snow will be over-estimated if these non-BC components are not accounted for

Post-wet-depositional processes are important! Most of the variability in snow particulate light absorption is driven by what’s happening between new snowfall events - dry deposition, sublimation, melting Overall findings Melt amplification: generally confined to the top few cm of the snow increases concentrations of BC/absorbing particles by up to a factor of about five . Scavenging fractions with melt-water: 10-30%

Idaho & Utah: dry deposition and in-snow processes increase the mixing ratio of : BC by up to an order of magnitude all light-absorbing particulates by up to 2 orders of magnitude

Spatial variability at a range of scales is considerably smaller than the temporal variations at a given site implications for the representativeness of field samples used in observation/model comparisons. Zhang et al., 2015, ACP Qian et al., 2014, Env. Res. Lett. Data model source study using study of BC-in-snow radiative forcing tagged emissions are being used to test/adjust models

Radiative forcing by BC in snow Publications: Wang, X., S. J. Doherty and J. Huang, Black carbon and other light-absorbing impurities in snow across Northern China, J. Geophys. Res. Atmos., 118 (3), 1471-1492, doi: 10.1029/2012JD018291, 2013.

Doherty. S. J., C. Dang, D. A. Hegg, R. Zhang and S. G. Warren, Black carbon and other light-absorbing particles in snow of central North America, J. Geophys. Res. Atmos., 119, doi:10.1002/2014JD022350, 2014.

Dang, C. and D. A. Hegg, Quantifying light absorption by organic carbon in Western North American snow by serial chemical extractions, J. Geophys. Res. Atmos., 119, 10,247-10,261, doi:10.1002/2014JD022156.

Zhang, R., H. Wang, D. A. Hegg, Y. Qian, S. J. Doherty, C. Dang, P.-L. Ma, P. J. Rasch, and Q. Fu, Quantifying sources of black carbon in Western North America using observationally based analysis and an emission tagging technique in the Community Atmosphere Model, Atmos. Chem. Phys., 15, 12805-12822, doi:10.5194/ acp-15-12805-2015, 2015.

Doherty, S. J., D. A. Hegg, P. K. Quinn, J. E. Johnson, J. P. Schwarz, C. Dang and S. G. Warren, Causes of variability in light absorption by particles in snow at sites in Idaho and Utah, J. Geophys. Res. Atmos., 121, doi:10.1002/2015JD024375, 2016.

+ other studies that used our data (e.g. Qian et al., 2014, Environ. Res. Lett.)