This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 129.2.24.164

This content was downloaded on 09/02/2016 at 19:30

Please note that terms and conditions apply.

Vegetation fires, absorbing aerosols and smoke plume characteristics in diverse biomass

burning regions of Asia

View the table of contents for this issue, or go to the journal homepage for more

2015 Environ. Res. Lett. 10 105003

(http://iopscience.iop.org/1748-9326/10/10/105003)

Home Search Collections Journals About Contact us My IOPscience

Environ. Res. Lett. 10 (2015) 105003 doi:10.1088/1748-9326/10/10/105003

LETTER

Vegetation fires, absorbing aerosols and smoke plumecharacteristics in diverse biomass burning regions of Asia

Krishna PrasadVadrevu1, Kristofer Lasko, LouisGiglio andChris JusticeDepartment ofGeographical Sciences, University ofMarylandCollege Park, College Park,MD20740,USA1 Author towhomany correspondence should be addressed.

E-mail: Krisvkp@umd.edu

Keywords: vegetation fires, absorbing aerosols, smoke plume heights

AbstractIn this study,we explored the relationships between the satellite-retrievedfire counts (FC),fire radiativepower (FRP) andaerosol indicesusingmulti-satellite datasets at a daily time-step covering tendifferentbiomassburning regions inAsia.Wefirst assessed the variations inMODIS-retrieved aerosol optical depths(AOD’s) in agriculture, forests, plantationandpeat landburning regions and thenusedMODISFCandFRP(hereafter FC/FRP) to explain the variations inAODcharacteristics.Results suggest that tropical broadleafforests inLaosburnmore intensively than theother vegetationfires. FC/FRP-AODcorrelations indifferentagricultural residueburning regionsdidnot exceed20%whereas in forest regions they reached40%.Tospecifically account for absorbing aerosols,weusedOzoneMonitoring Instrument-derived aerosolabsorptionoptical depth (AAOD) andUVaerosol index (UVAI). Results suggest relatively highAAODandUVAIvalues in forestfires comparedwithpeat andagriculturefires. Further, FC/FRPcould explain amaximumof29%and53%ofAAODvariations,whereasFC/FRPcould explain atmost 33%and51%ofthe variation in agricultural and forest biomass burning regions, respectively. Relatively,UVAIwas found tobe abetter indicator thanAODandAAODinboth agriculture and forest biomass burningplumes.Cloud–Aerosol Lidar and InfraredPathfinderSatelliteObservationsdata showedvertically elevated aerosol profilesgreater than3.2–5.3 kmaltitude in the forestfire plumes compared to2.2–3.9 kmand less than1 kminagriculture andpeat-landfires, respectively.We infer theneed to assimilate smokeplumeheightinformation for effective characterizationof pollutants fromdifferent sources.

1. Introduction

Biomass burning is an important source of aerosolsand greenhouse gas emissions in several regions of theworld including Asia (Seiler and Crutzen 1980). Thecausative factors of biomass burning vary by region.For example, in northeast India, Eastern Ghats, north-east Myanmar, Laos and Cambodia, biomass burninghas beenmainly attributed to slash andburn agriculture(Toky and Ramakrishnan 1983, Prasad et al 2002, 2008,Palm et al2013). InnorthwesternThailand, theMekongDelta in southern Vietnam, and the Punjab region ofIndia, most of the biomass burning is attributed toagricultural residues. Farmers burn crop residues afterharvest to control pests andweeds, improve soil fertilitythrough ash and to facilitate planting of new crops(Gadde et al 2009, Taylor 2010, Vadrevu et al 2011,Kharol et al 2012, Sahu and Sheel 2014). In contrast,forests in the peat land burning for oil palm plantations

is most common in Indonesia, Malaysia and Papau

New Guinea (Carlson et al 2013). Also, vegetation and

peatland fires in Southeast Asia have been attributed to

a combination of El Nino-induced droughts and

anthropogenic land-use changes (Langner et al 2007,Gaveau et al 2014). Several studies have shown that

aerosols and pollutants from biomass burning can be

transported long distances and persist for weeks to

months, impacting not only air quality but also

biogeochemical cycles, atmospheric chemistry,

weather, and climate (Radojevic 2003, Cristofanelli

et al 2014, Reddington et al 2014). In addition, biomass

burning pollutants can have significant health impacts

with increased respiratory ailments, eye irrigation,

medication use and exacerbated asthma (Laumbach

and Kipen 2012). It is therefore important to character-

ize the emissions from biomass burning sources more

accurately.

OPEN ACCESS

RECEIVED

28March 2015

REVISED

2 September 2015

ACCEPTED FOR PUBLICATION

8 September 2015

PUBLISHED

14October 2015

Content from this workmay be used under theterms of theCreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2015 IOPPublishing Ltd

One of the important indicators of aerosol amountsin the atmospheric column is aerosol optical depth(AOD), which is a measure of atmospheric extinctionthrough a vertical column of atmosphere (Carmichaelet al 2009). The aerosol components contributing toAOD include soot, sulfates, organics, dust, etc., andinclude both natural and anthropogenic sources (Bar-naba and Gobbi 2004). A number of ground-basedmea-surement stations of the AErosol RObotic NETwork(AERONET) exist in the world (Holben et al 1998). TheAERONET consists of CIMEL sun/sky radiometers cap-able of retrieving aerosol optical products at discretewavelengths ranging from 440nm (visible) to 1020 nm(near IR) (Eck et al 1999, Schuster et al 2006). The mea-surements from AERONET includes AOD, precipitablewater, fine and coarse mode AOD including fine modefraction, sky and surface radiance for bi-directional reflec-tance distribution functionDue to the large temporal andspatial variability in aerosol composition and abundance,satellite retrievals of AOD became more useful for char-acterizing aerosols in diverse regions of the world(Myhre 2009). For example, MODIS AOD is retrievedusing multiple channels onboard the Terra and Aquasatellites beginning in 2000 and 2002, with separate algo-rithms for oceans (Tanré et al 1997) and land (Kaufmanet al 1997). The AOD’s are derived by the inversion of theMODIS-observed reflectance using pre-computed radia-tive transfer look-up tables based on aerosol models(Remer et al2002).

Satellite observations such as AOD can be a powerfultool for monitoring of atmospheric pollution if they canbe related to theunderlying emission sources, especially inreal-time. Specific to biomass burning emissions, activefire detection and radiative power products are beingusedin emissions estimation in real-time such as through theGlobal Fire Assimilation System. The system is con-tinuously refined to account for fire-emission amounts(Kaiser et al 2012). To make such operational systemsrobust, there is a need to evaluate and refine fire–aerosolrelationships inmultiple regions. Specific to Asia, most ofthe biomass burning aerosols are often mixed from fossilfuel combustion and dust emissions (Kaskaoutiset al 2009, Bucci et al 2014, Mishra et al 2014). In such acontext, an important question to address is ‘how muchof the AOD increase is due to biomass burning and howwell do satellitefire retrievals explainAODvariation?’

In this study,wecharacterizefire–aerosol relationshipsinAsia.Wefirst assessed thevariation inMODIS-retrievedAOD’s in diverse biomass burning regions pertaining toagriculture, forests andplantationburning. In addition,weaddressed the following questions: How well do satellite-derived fire products (fire counts (FC) and fire radiativepower (FRP) (hereafter FC/FRP) correlate with differentaerosol indices such as AOD, UV aerosol index (UVAI)and aerosol absorption optical depth (AAOD)? Is FRP abetter indicator than FC in relating to AOD, AAOD andUVAI? What are the typical smoke plume heights in thebiomass burning regions of Asia? How do smoke plumeheights influence fire-AOD, AAOD, UVAI relationships?

How do the correlations vary over different burningregions (i.e. agriculture versus forest versus peatland) dur-ing the peak biomass burning months? We addressedthese questions usingMODIS, OzoneMonitoring Instru-ment (OMI) andCloud–Aerosol Lidar and Infrared Path-finderSatelliteObservation (CALIPSO)datasets.

2.Datasets andmethods

We selected ten different biomass-burning windows inAsia to characterize fire–aerosol relationships (figure 1).Each window was comprised of 3×3 one-degreeresolution cells (each cell with 111.3 sq.km). The domi-nant vegetation typeburned in eachwindowand thepeakmonths of biomass burning are given in table 1 andfigure 1. Punjab located in the northwestern region ofIndia is dominated by rice–wheat crop rotations whereagricultural residue burning is prevalent (Vadrevuet al 2011, 2013). Eastern Ghats are a range of discontin-uous mountains situated on the east coast of India. Theyrun fromWest Bengal state in the north, throughOdishaand Andhra Pradesh to Tamil Nadu in the south passingsome parts of Karnataka State. Biomass burning inEastern Ghats is mostly due to slash and burn. NortheastIndia comprises of seven-different states in India whichincludes Assam, Arunachal Pradesh, Manipur, Megha-laya, Mizoram, Tripura and Nagaland. Slash and burnagriculture ismost common innortheast India.Theotherregions include north-western Thailand with dominantrice–maize crop burning, Laos and Eastern Myanmarwith tropical broadleaf forest burning, southern Vietnam(Mekong delta) with rice residue burning and Indonesiawith forest/peat landfires.

We used the MERIS Globcover land cover dataset(version 2.3) for inferring the land cover types. The dataset contains 22 separate land cover classes created fromcloud-freemosaics ofMERIS surface reflectance data andvegetation indices at a 300m spatial resolution (Bicheronet al2008). For characterizing thefire activity andFRP, thedominant land cover category was inferred to each bio-mass burning window based on a majority filter and thedominant vegetation typeburntbasedon the literature.

For characterizing fires and FRP, we used daily activefire detections from a combination of theMODIS instru-ments onboard the Aqua and Terra satellites. The twoMODIS sun-synchronous, polar-orbiting satellites passover the Equator at approximately 10:30 a.m./p.m.(Terra) and 1:30 p.m./a.m. (Aqua)with a revisit time of 1to 2 days. TheMODIS Advanced Processing System pro-cesses the resulting data using the enhanced contextualfire detection algorithm (Giglio et al 2003) combined intothe Collection 5 Active Fire product. For this study, weanalyzed the daily FC and FRP data for the peak biomassburning periods from 2005 to 2012 and data with morethan 95% confidence level (Giglio 2009). Correspondingto the similar time-period, we used daily MODIS AODproducts (MOD08_D3.005 and MYD08_D3.005) at550 nm. To assess the statistical nature of AOD data

2

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

during the peak biomass burning months in differentregions of Asia, we performed a frequency analysis. TheMODIS AOD data were fitted to a Gaussian distributionto infer location and scale parameters as

y yA

2e ,

x x

02

c2

2

( )s p

= + s-

-

/

where A is the amplitude, xc is the center of thepeak amplitude, and s (sigma) is the width at half peakamplitude.

The AAOD is a columnar measure of concentra-tion of near-UV absorbing aerosol particles such assmoke andmineral dust and is retrieved from theOMI(Bucsela et al 2008). We used the AAOD daily productat 500 nm (OMAERUVd.003). In addition, we alsoused UV aerosol index (UVAI) that detects the pre-sence of UV-absorbing aerosols such as dust and soot.

UVAI is based on a spectral contrast method in a UVregion where the ozone absorption is very small. It isthe difference between the observations and modelcalculations of absorbing and non-absorbing spectralradiance ratios. ForOMI, AI is defined as

I I

I I

AI 100 log

log .

10 360 331measured

10 360 331calculated

( )( )

⎡⎣⎤⎦

=

-

Positive values of AI generally represent absorbingaerosols (dust and smoke) while small or negativevalues represent non-absorbing aerosols (sulfate, sea-salt) and clouds (Torres et al 2007). We specificallyused the daily product (OMTO3d.003) to infer UVAIvariations. Both the AAOD and UVAI variations wereassessed using histograms with mean and standarddeviation. To address fire-AOD, AAOD and UVAI

Figure 1. Study area locationmapwith biomass burning regional windows.

Table 1.Biomass burning regionswith peak biomass burningmonths and the type of biomass burnt.

Site Peak biomass burningmonths Dominant vegetation burnt

Thailand February–April Rice–maize crop residues

SouthernVietnam February–April Rice residues

Punjab, India-summer March–May Wheat residues

Punjab, India-winter October–December Rice residues

Riau, Indonesia July–September Forest/peat landfiremix

Kalimantan, Indonesia July–September Peat landfiremix

EasternGhats, India February-April Tropicalmixed deciduous forest

Northeast India March–May Tropical evergreen forest

EasternMyanmar February–April Broad leaf deciduous forest

Laos February–April Broadleaf forest

3

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

relationships, we used linear regression with 95% con-fidence interval bands.

To attribute thedifferences infire-AODrelationships,we used the CALIPSO data. The vertical distribution ofaerosol information is inferred through the analysis ofbackscatter measurements provided by the Cloud–Aero-sol Lidar with Orthogonal Polarization (CALIOP) instru-ment onboard the CALIPSO satellite of the NASAA-Train (Winker 2007, Hunt et al 2009). The CALIPSOproduct is highly useful in discriminating aerosol layersfrom clouds (Liu et al 2009), categorizing aerosol layers asone of six subtypes (dust, marine, smoke, polluted dust,polluted continental, and clean continental; Omaret al 2009), and for estimating the optical depth of eachlayer detected (Vaughan et al 2004). In this study, we spe-cifically used theCAL_LID_L1-ValStage1-V3-30datasets;averaged three to fourmostlynight-timepasses during thepeak biomass burning periods for each window as theCALIPSO provides information every 16 days (Win-ker 2007).Weused the total attenuated backscatter coeffi-cient at 532 nm and smoke altitude information forcharacterizing aerosol concentrations during the biomassburningmonths fordifferent regionalwindows (figure1).

3. Results

3.1. FC variations (2005–2012)The sumof FC for the peak biomass burningmonths in atypical 3×3 one degree resolution window for differentbiomass burning regions are shown in box and whisker

plotswith thebottomand topof thebox representingfirstand third quartiles, the band inside the box with medianvalues, and the tails representing the minimum to themaximum FC (figures 2(a) and (b)). In agriculturalbiomass burning regions, highest FC were found duringthe Punjabwinter season (1350FC) followed byThailand,Punjab summer and Southern Vietnam. The median FCvalues were higher for Punjab winter data compared tothe others (figure 2(a)). Of the different regions, Laoswiththe broadleaf forest burning had the highest FC (2596)and lowest FC in Riau, Indonesia including the medianvalues (figure2(b)).

3.2. FRP variationsAmongst the agricultural biomass burning regions,maximum FRP was found for Thailand (318.3 MW)followed by Punjab during summer (figure 2(c)).Among the forest, plantation and peat land fires, Laoshad the highest FRP (630.7 MW) and lowest forKalimantan peat land fires (338.8 MW) (figure 2(d)).These results clearly suggest more intense biomassburning from tropical broadleaf forests (Laos) thanfromagriculture and peat land fires.

3.3. AODvariationsBiomass burning activities are expected to increase AOD’scompared to the background (Badarinath et al 2007, 2009,Eck et al 2009). Figures 3(a)–(j) summarizes the MODIS-retrieved AODs based on the fitted Gaussian model. Forthe normal distribution, the location and scale parameters

Figure 2. (a)–(d)Variations infire counts (FC) (a), (b) and FRP (c), (d) in different biomass burning regions. The box andwhiskerplots showminimum tomaximumvalues in FC and FRP, respectively. Figures 2(a) and (b) refer to agricultural burning regions;whereas except for Riau andKalimantan (forest/peat landmix fires), the other regions in (b) and (d) refer to forest biomass burningregions. Refer to table 1 for the peak biomass burningmonths and dominant vegetation type burned.

4

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

correspond to the mean (μ) (given as Mu in figures) andstandard deviation (sigma). For example, in the entiredataset, the mean AOD was relatively high for Laos(μ=0.75) broadleaf forest burning (figure 3(h))

compared to the other vegetation types including agricul-ture categories 3(a)–(d). Similarly, Riau province, Indone-sia with the forest/peat landmix fires (figure 3(i)) had thehighest sigma values (0.592) compared to the other forest

Figure 3. (a)–(i)AODFrequency plotswith normal distribution for different biomass burningwindows.Mean and sigma values arealso shown in the plots. (a)Punjab, summer, India; (b)Punjab, winter, India; (c) SouthernVietnam; (d)Thailand; (e)EasternGhats,India; (f)Northeast India; (g)EasternMyanmar; (h) Laos; (i)Riau, Indonesia; (j)Kalimantan, Indonesia.

5

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

burning regions. Among the forest types, tropical mixeddeciduous forest in Eastern Ghats (figure 3(e)) had thelowest mean AOD (0.361) and sigma (0.178). Further,among the crop residue burning regions, Punjab, Indiahad the highestmean both in summer (figure 3(a)) aswellaswinter (figure 3(b)) (μ=0.621 and 0.633, respectively),whereas, southern Vietnam with rice residue burning(figure 3(c)) had the lowest mean AOD (0.441). We alsonote that forest/peat land mix biomass burning regionshad relatively higher standard deviations indicating highervariability inAODcompared to the agricultural regions.

3.4. AAODvariationsAAOD variations for different sites are shown ashistograms with mean and standard deviation(figures 4(a) and (b)). Among the agricultural biomassburning regions, Punjab during the summer season hadthe highest mean AAOD (0.075) and lowest in SouthernVietnam (0.038) (figure 4(a)). The standard deviation inAAOD values was similar (0.044) for Punjab, SouthernVietnam, and Thailand and lowest during the Punjabwinter season (0.036) (figure 4(b)). Among the forest,plantation and peat land fires, Laos and Myanmar withforest biomass burning had the highest AAOD (0.117)andEasternGhats the lowest (figure4(b)).

3.5. UVAI variationsSimilar to AAOD, Punjab during the summer seasonshowed highest mean UVAI (1.83) (figure 4(c)).Among the forested biomass burning regions, Laoshad the highest mean UVAI (1.58) and the lowest forthe peat-landfires inKalimantan (0.87) (figure 4(d)).

3.6. FC-AODandFRP-AODcorrelationsFC-AOD correlations (figures A1(a)–(i) in appendix)in different agricultural regions did not exceed 20%(figures A1(a)–(d) in appendix). Highest FC-AODcorrelations were found for Thailand and Punjabwinter burning (both with r2=0.20 (figures A1(d),(b)) and lowest for Punjab-summer burning(r2=0.08) (figure A1(a)). With the exception insouthern Vietnam, FRP-AOD correlations were con-sistently weak in different agricultural systems and lessthan FC-AOD relationships. FC-AOD correlations inforest, plantation and peat land fires were relativelyhigher than the agricultural systems with highestcorrelation (r2=0.34) in Myanmar A1(g) and lowestin Riau province (r2=0.02) (figure A1(i)). FRP-AODcorrelations were relatively higher for Myanmar(r2=0.40) (figure A3(g)) compared to the others.

3.7. FC-AAODand FRP-AAODCorrelationsOur results indicate comparatively lower correlationsbetween FC-AAOD and FRP-AAOD in agriculturalsystems than in forest systems (figures A1–A2(a)–(i),appendix). In agricultural systems, the highest FC-AAOD correlation was found for Thailand (r2=0.29)(figure A1(d)) and lowest for southern Vietnam(r2=0.078) (figure A1(c)). Further, FC-AAOD andFRP-AAOD correlations were almost similar in agri-cultural systems. Among the forest biomass burningregions, the highest FC-AAOD correlation was in Laos(r2=0.53) (figure A1(h)) and lowest in Riau province(r2=0.18) (figure A1(i)).

Figure 4. (a)–(d)Variations inAAOD (a), (b) andUVAI (c), (d) in different biomass burning regions. The column bar graphs showmeanwith standard deviation.

6

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

3.8. FC-UVAI and FRP-UVAI correlationsThese correlations in agricultural systems were muchstronger than FC-AOD and FC-AAOD correlations(figures A1–A4(a)–(j), appendix). Amongst the agricul-tural systems, FC-UVAI correlation was highest in Thai-land (0.33) (figures A2(d)) and lowest during the Punjab-summer season (r2=0.14) (figure A2(a)). FRP-UVAIcorrelations were similar to FC-UVAI correlations inagricultural systems. Further, FC-UVAI correlations inforest, plantation andpeat landfireswere relatively higherthan the agricultural systems with the highest correlationin Laos (r2=0.51) (figure A2(h)) and lowest in Riau

province (r2=0.20) (figure A2(i)). Also, FRP-UVAIcorrelations were relatively stronger for forests thanagriculture andpeatlandfires.

In summary, the correlation analysis suggested the fol-lowing: (a) The FC/FRP versus AOD, AAOD and UVAIcorrelations were stronger for forest, plantation and peatland fires than the agriculturalfires; (b)Among the aerosolindices, UVAI was more strongly correlated with FC orFRP followedbyAAODandAODinboth forest, peat landand plantation/peat land mix fires than the agriculturalfires; (c)Our results indicate that that the sum of FRP wasbetter correlatedwithUVAI thanAODorAAOD in forest

Figure 5. (a)–(c) Smoke plume characteristics in different biomass burning regions retrieved fromCALIPSOdata. Specific biomassburning regions are highlighted in blue elliptic circles. (a) Laos with smoke plume altitude (∼5.3 km) representing broad leaf forestburning; (b)Kalimantan, Indonesia with smoke plume altitude (800 m) representing peat-land fires; (c)Punjab, India, smoke plumeheights (2.3 km) during summer agricultural residue burning. Polluted dust aerosols above 5 km can also be seen in the blue ellipticcircle in orange color (in contrast to smoke in black color).

7

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

biomass burning regions. Further, we found UVAI as bet-ter indicator thanAODorAAODandcorrelatingwellwithFRPthanFCin forest, plantationandpeat landfires.

3.9. CALIPSOobservationsIn table 2, CALIPSO-derived 532 nm backscatter valuesas well as smoke plume heights for different biomassburningregionsandaerosol sub-typeswithplumeheightsfor sample regions in figures 5(a)–(c). Vertical profiles ofaerosol concentrations showed maximum concentra-tions over the forested regions compared to the agricul-tural biomass burning sites (table 2). Specific to theagricultural residueburning sites, aerosol loading is foundtobe highest forThailand (3×10−2/sr/km) followedbyPunjab-winter season (4.5×10−3/sr/km) and lowestfor southern Vietnam. Aerosols reaching the highestaltitudehavebeennoted forThailand (3.9 km) and lowestfor southern Vietnam (2.2 km). Eastern Ghats, India aswell as Kalimantan and Riau province, Indonesia hadrelatively low smoke plume altitudes than other forestbiomassburning sites.

4.Discussion

The above results suggest that a broad distinction can bemade between the aerosol properties and smoke plumeheights of agricultural versus forest biomass burningregions. Most of the forest biomass burning regions hadrelatively higher median FC and FRP. Results clearlysuggest that tropical broadleaf forests in Laos burn moreintensively (higher FRP) than the other forest types, peatlands as well as agricultural regions. Higher FRP fromforests may be attributed to relatively higher biomass perunit area compared to agriculture and peatlands. How-ever, more verification is needed to delineate FRPvariations in the field. As in several other studies(Ramanathan et al 2001, Eck et al 2009, Lin et al 2014), wenoted increase in AOD during the biomass burningmonths. The mean AOD during the peak biomassburning periods in the Laos exceeded 0.75. Similarly inthe agricultural biomass burning regionsmean AODwasgreater than 0.60 in Punjab summer as well as during

winter residue burning. Although AOD is a goodindicator of overall air pollution, increase inAODmay beattributed to a variety of other such as dust aerosols (Kimet al2007,Vadrevu et al2011). Further,MODISAODcanonly be retrieved under clear-sky conditions and underpartially cloudy skies the probability of subpixel cloudcontamination canbe larger (Xia et al2013).

The main aerosol released from biomass burning thatcauses large variations in the atmospheric chemistry andradiation budget is black carbon, which is the opticallyabsorbing part of the carbonaceous aerosols (Saha andDespiau2009). Inparticular,OMIAAODis an indicatorofabsorbing carbonaceous aerosols resulting from biomassburning activity (Torres et al 2010). The AAOD values foragriculturalfires varied from0.03–0.07 to 0.08–0.11 for theforest/plantation and peat land mixed fires. These valuesclearly suggest biomass burning from forest fires releasemuch more absorbing aerosols than the agricultural fires.The AAOD values reported for forest biomass burning inour studywere comparatively lower than theAAODvaluesreported for fire plumes in South America and CentralAfrica. For example,Torres et al (2010)during thefive-yearanalysis period reported thepeakmonthlyAAODvalues inthe range of 0.08–0.15 during September in the SouthAmerica. They also reported peak monthly AAOD valuesbetween 0.08 and 0.12 during August in Central Africa.Although we observed relatively higher FC-AAOD andFRP-AAOD correlations for the forest/plantation peatland mix fires than the agricultural fires, the correlationwithin the sites betweenFC andFRPwithAAODwere notmuch different. Further the higher correlations observedbetween FC/FRP-AAOD compared to FC/FRP-AODsuggests AAOD as a better indicator of absorbing aerosolsfor forest andplantationpeat landmixfires.

Specific to the OMI-UVAI, earlier, Torres et al(1998, 2007, 2010), Hsu et al (1999), and Ginoux andTorres (2003) noted an important sensitivity of OMI toUV absorbing aerosols as a function of smoke plumeheights. Also, Torres et al (1998, 2012) showed that themagnitude of the positive AI depends on the AAOD andheight of the aerosol layers. They also showed that AI sig-nal is meaningful above 0.5 and AI signal is amplifiedwhen the absorbing layer lies above the clouds. In essence,

Table 2. SmokeplumeandPolluteddust altitudes fordifferentbiomassburning regions retrievedusingCALIOPdata.Highest smokeplumealtitudes (5.3 km) canbe seen for Laos tropical broad-leaf forest biomass burning and lowest forKalimantan, Indonesiapeatlandfires (800 m). Relativelyhigherbackscatter valueswerenoted for forest biomass burning sites compared to the agriculture andpeat landfires.

Site Smoke altitude (Km)Polluted dust alti-

tude (Km)CALIPSO 532 nm total attenuated backscatter,

km−1 sr−1

Thailand 3.9 2.8 3×10−2

SouthernVietnam 1.3 2.2 4×10−3

Punjab, India-summer 2.3 5.2 2.5×10−3

Punjab, India-winter 2.5 2.9 4.5×10−3

Riau, Indonesia 3.2 3 5.0×10−3

Kalimantan, Indonesia 800 m 1 4.5×10−3

EasternGhats, India 3.2 3.6 5.0×10−3

Northeast India 4.8 4.9 7.0×10−2

EasternMyanmar 4.9 2.5 5.5×10−3

Laos 5.3 3.5 6.5×10−2

8

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

OMI AI can be retrieved whenmixed with clouds but itssensitivity increases significantly when the absorbingaerosol layers are located above clouds (Xia et al 2013). Inaddition, Badarinath et al (2009) and Guan et al (2010)have shown increasing AI greater than 1.5 in biomassburning plumes. Similar to these studies, we foundUVAIto be a good indicator of enhanced aerosols frombiomassburning, as the values were significantly high varyingfrom 1.0 to 1.8 in agricultural biomass burning plumesand 0.8–1.58 UVAI in forest and plantation/peat landmix fires. Consistently strong correlations (0.40–0.51)between FC-UVAI and (0.40–0.52) between FC-FRPwere observed among the forest fires in Eastern Ghats,Northeast India, Myanmar and Laos compared to FC/FRP-AOD and may be attributed to release of smokeplumes above clouds that could be effectively detected byOMI data (Torres et al 2010) compared to MODIS. Incontrast, the relatively weak FC/FRP-UVAI correlationsin peat and agricultural biomass burning regions may beattributed to low smoke plumes heights. To test the rela-tive differences in smoke plume heights in agriculturaland forest and plantation peat land mix fires, we eval-uated theCALIPSOdata.

Large amounts of smoke aerosol can be injected intothe atmosphere as a result of biomass burning. Literaturereview suggest that, unlike clouds, thin smoke plumeshave a weak 532 nm total attenuated backscatter signaland dense plumes may have a 532 nm backscatter signalsimilar to clouds (Winker 2007, Omar et al 2009). In ourcase, the overall CALIPSO backscatter values at 532 nmwere considerably higher for the forest biomass burningsites than the agricultural sites (figures 5(a)–(c)).

For example, Mishra et al (2014) reported2.32–3.58×10−3 sr km−1 for the Indo-Ganges regionclose to our reported values of 2.5×10−3 sr km−1 forPunjab during the summer biomass burning region.In the Punjab winter biomass burning season, weobserved higher backscatter values compared to thesummer. This may be due to relatively low dust signalduring winter compared to the summer. High smokeplume altitudes for the dust aerosols can be seen infigure 5(c) for the Punjab summer season. We alsonote comparable smoke plumeheights for Punjab duringboth summer and winter. Overall, the smoke plumeheights for agricultural regions were considerably lowerthan the forest biomass burning sites. Further, in Kali-mantan, Indonesia lowest smoke plume heights wereobserved with less than 1 km. Similar results were repor-ted by Tosca et al in the range of 709±14m on Borneoand749±24monSumatra and attribute it to low fireintensities from smoldering peat-land fires comparedto the large boreal crown fire plumes of Alaska, or thehigher-intensity grass fires of Australia (Kahn et al2008, Mims et al 2010). Further, in our study, UVAIvalues varied from 0.87 to 1.87 in the Kalimantanregion compared to 1–02–4.2 UVAI values observedfor other forest biomass burning regions. These dif-ferences clearly suggest increased pyro-convection ofaerosols over the forest biomass burning regions than

the agriculture as well as peat-land fires thus impactingFC/FRP-AOD, AAOD andUVAI correlations. As UVAIis sensitive to aerosols that aloft above the clouds, it can beused effectively in conjunction with AAOD and FC/FRPto relate to biomass burning aerosols in forest biomassburning regions.Wealso found smoke altitudes as key forrelating FC or FRPwith AOD, AAOD, UVAI in differentbiomass burning regions of Asia. More specifically, relat-ing FC/FRP to UVAI may be more justified than usingAODalone as theUVAI is a robust indicator of absorbingaerosols mainly from biomass burning compared toAOD which is mixed signal from several other aerosolsapart from biomass burning. For resolving inter-annualdifferences in FC/FRP and aerosol indices, more in-depth studies are needed. Variations in FC/FRP and cor-relations with the other atmospheric satellite data pro-ducts suggested distinct patterns among differentbiomass burning regions and the significance of UVbased aerosol index. We also infer the need to assimilatesmoke plume height data such as from CALIPSO foroperationalmonitoring of pollutants fromdifferent sour-ces includingbiomassburning to reduceuncertainties.

5. Conclusions

Fire–aerosol characteristics were studied using dailymulti-satellite data during the peak biomass burningperiods in different regions of Asia. MODIS FC as well asFRPproductswere useful in characterizingfire events andintensities. We found relatively lower FRP’s over theagricultural and peat land fire regions compared to forestbiomass burning regions. In the study, we also tested therelationship between FC/FRP and various aerosol pro-ducts suchasAOD,AAODandUVAI inbiomassburningregions. Of the different products, we found UVAI as abetter indicatorofbiomassburningpollutionandstronglyrelating to FC/FRP in forest biomass burning than theagricultural biomass burning regions. We used theCALIPSO data to infer poor/strong correlations of FC/FRP in agriculture and forest biomass burning regions.Results suggested significantly higher CALIPSO back-scatter as well as smoke plume heights for forest biomassburning regions than agriculture and peat-land fires. Theresults highlight FC/FRP-AOD, AAOD and UVAI rela-tionships in different biomass burning regions in additionto smoke plume characteristics useful to address biomassburningpollutiononatmosphere andclimate.

Acknowledgments

We thank product developers of MODIS, MERIS,OMI and CALIPSO for free sharing of products. Thisresearchwas supported byNASAgrantNNX10AU77G.

Appendix

9

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

Figure A1–A2. (a)–(j) Scatterplots between sumof FC-AOD, Sumof FC-AAODA1(a)–(j) and sumof FC-UVAI relationships A2(a)–(j). R-square linearfit and regression bands at 95% confidence interval are also shown. A1–A2(a)Punjab, summer, India; A1–A2(b)Punjab, winter, India; A1–A2(c) SouthernVietnam; A1–A2(d)Thailand; A1–A2(e)EasternGhats, India; A1–A2(f)Northeast India;A1–A2(g)EasternMyanmar; A1–A2(h) Laos; A1–A2(i)Riau, Indonesia; A1–A2(a), (j)Kalimantan, Indonesia.

10

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

Figure A1–A2. (Continued.)

11

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

Figure A3–A4. (a)–(j) Scatterplots between sumof FRP-AOD, sumof FRP-AAODA3(a)–(j) and sumof FRP-UVAI relationshipsA4(a)–(j). R-square linearfit and regression bands at 95% confidence interval are also shown. A3–A4(a)Punjab, summer, India; A3–A4(b)Punjab, winter, India; A3–A4(c) SouthernVietnam; A3–A4(d)Thailand; A3–A4(e)EasternGhats, India; A3–A4(f)NortheastIndia; A3–A4(g)EasternMyanmar; A3–A4(h).Laos; A3–A4(i)Riau, Indonesia; A3–A4(j)Kalimantan, Indonesia.

12

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

Figure A3–A4. (Continued.)

13

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

References

BadarinathKV S et al 2007Multiyear ground-based and satelliteobservations of aerosol properties over a tropical urban areain IndiaAtmos. Sci. Lett. 8 7–13

BadarinathKV S et al 2009Analysis of aerosol and carbonmonoxide characteristics over Arabian Sea during cropresidue burning period in the Indo-Gangetic Plains usingmulti-satellite remote sensing datasets J. Atmos. Sol.—Terr.Phys. 71 1267–76

Barnaba F andGobbiGP 2004Aerosol seasonal variability over theMediterranean region and relative impact ofmaritime,continental and Saharan dust particles over the basin fromMODIS data in the year 2001Atmos. Chem. Phys. 4 2367–91

Bicheron P et al 2008Geo-location assessment ofMERISGlobCoverortho-rectified productsProc. 2ndMERIS-AATSRWorkshop(Frascati, Italy, 22–26 September 2008) SP-666

Bucci S, CagnazzoC,Cairo F,Di Liberto L and Fierli F 2014Aerosolvariability and atmospheric transport in theHimalayanregion fromCALIOP2007–2010 observationsAtmos. Chem.Phys. 14 4369–81

Bucsela E J et al 2008Comparison of troposphericNO2 from in situaircraftmeasurements with near real-time and standardproduct data fromOMI J. Geophys. Res. 113 2156–202

CarlsonKM,CurranLM,AsnerGP,PittmanAM,Trigg SNandAdeney JM2013Carbonemissions from forest conversionbyKalimantanoil palmplantationsNat.Clim.Change3283–7

Carmichael GR et al 2009Asian aerosols: current and year 2030distributions and implications to human health and regionalclimate change Environ. Sci. Technol. 43 5811–7

Cristofanelli P et al 2014Transport of short-lived climate forcers/pollutants (SLCF/P) to theHimalayas during the SouthAsiansummermonsoon onset Environ. Res. Lett. 9 084005

EckT F et al 1999Wavelength dependence of the optical depth ofbiomass burning, urban, and desert dust aerosols J. Geophys.Res. 104 31333–49

EckT F et al 2009Optical properties of boreal region biomassburning aerosols in central Alaska and seasonal variation ofaerosol optical depth at anArctic coastal site J. Geophys. Res.114D11201

Gadde B,MenkeC andWassmannR 2009Rice straw as a renewableenergy source in India, Thailand, and the Philippines: overallpotential and limitations for energy contribution andgreenhouse gasmitigationBiomass and Bioenergy 33 1532–46

GaveauDL et al 2014Major atmospheric emissions frompeat firesin Southeast Asia during non-drought years: evidence fromthe 2013 Sumatran fires Sci. Rep. 4 6112

Giglio L et al 2003An enhanced contextualfire detection algorithmforMODISRemote Sens. Environ. 87 273–82

GiglioL et al2009Anactive-fire basedburnedareamapping algorithmfor theMODIS sensorRemote Sens. Environ.113408–20

Ginoux P andTorresO 2003 Empirical TOMS index for dustaerosol: applications tomodel validation and sourcecharacterization J. Geophys. Res. 108 4534

GuanH et al 2010Amulti-decadal history of biomass burningplume heights identified using aerosol indexmeasurementsAtmos. Chem. Phys. 10 6461–9

Holben BN et al 1998AERONET—A federated instrument networkand data archive for aerosol characterizationRemote Sens.Environ. 66 1–16

HsuNC et al 1999Comparisons of the TOMS aerosol indexwithSun-photometer aerosol optical thickness: results andapplications J. Geophys. Res. 104 6269–79

HuntWH et al 2009CALIPSO lidar description and performanceassessment J. Atmos. Ocean. Technol. 26 1214–28

KahnRA et al 2008Wildfire smoke injection heights: twoperspectives from spaceGeophys. Res. Lett. 35 L04809

Kaiser JW et al 2012 Biomass burning emissions estimatedwith aglobalfire assimilation systembased on observed fireradiative powerBiogeosci. 9 527–54

KaskaoutisDG et al 2009Variations in the aerosol optical propertiesand types over the tropical urban site ofHyderabad, IndiaJ. Geophys. Res. 114D22204

KaufmanYJ et al 1997Operational remote sensing of troposphericaerosol over land fromEOSmoderate resolution imagingspectroradiometer J. Geophys. Res. 102 17051–67

Kharol SK et al 2012 Black carbon aerosol variations over Patialacity, Punjab, India—a study during agriculture crop residueburning period using groundmeasurements and satellite dataJ. Atmos. Sol.—Terr. Phys. 84 45–51

KimS-W et al 2007 Seasonal andmonthly variations of columnaraerosol optical properties over east Asia determined frommulti-yearMODIS, LIDAR, andAERONET sun/skyradiometermeasurementsAtmos. Environ. 41 1634–51

Langner A,Miettinen J and Siegert F 2007 Land cover change2002–2005 in Borneo and the role offire derived fromMODIS imageryGlob.Change Biol. 13 2329–40

LaumbachR J andKipenHM2012Respiratory health effects of airpollution: update on biomass smoke and traffic pollution J.Allergy Clin. Immunology 129 3–11

LinNH et al 2014 Interactions between biomass-burning aerosolsand clouds over Southeast Asia: current status, challenges,and perspectives Environ. Pollut. 195 292–307

Liu Z et al 2009TheCALIPSO lidar cloud and aerosoldiscrimination: Version 2 algorithm and initial assessment ofperformance J. Atmos. Ocean. Technol. 26 1198–213

Mims SR et al 2010MISR stereo heights of grassland fire smokeplumes in Australia IEEETransactions onGeosci. RemoteSens. 48 25–5

Mishra AT, Shibata T and Srivastava A 2014 Synergistic approachfor the aerosolmonitoring and identification of types overIndo-Gangetic Basin in Pre-Monsoon SeasonAerosol AirQual. Res. 14 767–82

MyhreG2009Consistency between satellite-derived andmodeledestimates of the direct aerosol effect Science 325 187–90

OmarAH et al 2009TheCALIPSO automated aerosol classificationand lidar ratio selection algorithm J. Atmos. Ocean. Technol.26 1994–2014

PalmCA et al 2013 Slash-and-Burn Agriculture: The Search forAlternatives (NewYork: ColumbiaUniversity Press)

PrasadVK, BadarinathKVS and EaturuA 2008 Biophysical andanthropogenic controls of forestfires in theDeccan Plateau,India J. Environ.Manage. 86 1–13

PrasadVK et al 2002 Biomass burning and related trace gasemissions from tropical dry deciduous forests of India: astudy usingDMSP-OLS data and ground-basedmeasurements Int. J. Remote Sens. 23 2837–51

RadojevicM2003Chemistry of forestfires and regional hazewith emphasis on Southeast Asia Pure Appl. Geophys. 160157–87

RamanathanVC et al 2001Aerosols, climate, and the hydrologicalcycle Science 294 2119–24

ReddingtonCL et al 2014Contribution of vegetation and peat firesto particulate air pollution in Southeast AsiaEnviron. Res.Lett. 9 094006

Remer LA et al 2002Validation ofMODIS aerosol retrieval overoceanGeophys. Res. Lett. 29 1618

SahaA andDespiau S 2009 Seasonal and diurnal variations of blackcarbon aerosols over aMediterranean coastal zoneAtmos.Res. 92 27–41

Sahu LK and Sheel V 2014 Spatio-temporal variation of biomassburning sources over South and Southeast Asia J. Atmos.Chem. 71 1–19

SchusterGL,DubovickOandHolbenBN2006Angstromexponent andbimodal aerosol sizedistributions J.Geophys.Res.111D07207

SeilerW andCrutzen P J 1980 Estimates of gross and net fluxes ofcarbon between the biosphere and the atmosphere frombiomass burningClim. Change 2 207–47

TanréD et al 1997Remote sensing of aerosol properties over oceansusing theMODIS/EOS spectral radiances J. Geophys. Res. 10216971–88

TaylorD 2010 Biomass burning, humans and climate change inSoutheast AsiaBiodiversity Conservation 19 1025–42

TokyOP andRamakrishnan PS 1983 Secondary successionfollowing slash and burn agriculture inNorth-Eastern India:I. biomass, litterfall and productivity J. Ecology 71 735–45

14

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al

TorresO et al 1998Derivation of aerosol properties from satellitemeasurements of backscattered ultraviolet radiation:theoretical basis J. Geophys. Res. Atmos. 103 17099e17110

TorresO et al 2007Aerosols and surfaceUVproducts fromozonemonitoring instrument observations: an overview J. Geophys.Res. 112D24S47

TorresO et al 2010OMI andMODIS observations of the anomalous2008–2009 SouthernHemisphere biomass burning seasonsAtmos. Chem. Phys. 10 3505–13

TorresO et al 2012Retrieval of aerosol optical depth above cloudsfromOMI observations: sensitivity analysis and case studies J.Atmos. Sci. 69 1037–53

VadrevuKP et al 2011MODIS derivedfire characteristics andaerosol optical depth variations during the agricultural

residue burning season, north IndiaEnviron. Pollut. 1591560–9

VadrevuKP et al 2013Hotspot analysis of vegetation fires andintensity in the Indian region IEEE J. Sel. Top. Appl. EarthObs. Remote Sens. 6 224–38

VaughanMA et al 2004 Fully automated analysis of space-basedlidar data: an overview of theCALIPSO retrieval algorithmsand data productsProc. SPIE 5575 16–30

WinkerDM2007 Initial performance assessment of CALIOPGeophys. Res. Lett. 34 19803

Xia X, ZongX and Sun L 2013 Exceptionally active agriculturalfire season inmid-easternChina in June 2012 and its impacton the atmospheric environment J. Geophys. Res. 1189889–900

15

Environ. Res. Lett. 10 (2015) 105003 KPVadrevu et al