Aerosols

The bushfire (wildfire) event of the 2019/2020 southeastern Australian summer has been described as unprecedented. It was unprecedented because of the extent of the area burned and the ferocity at which burning occurred. Between September 2019 and February 2020, 12 million hectares (120 000 square km) of land across the eastern coasts of Australia burned. The fires led to the direct loss of 34 lives, the destruction of 3 500 houses and significant wildlife and habitat losses.

Smoke from fires severely reduced air quality across the entire south-eastern region of Australia, from Queensland in the north to Tasmania in the south. Long-range transport of smoke was observed to reach New Zealand on several occasions. Figure 1 includes an image from the satellite Himawari-8 on 21 December 2019 showing the locations of fires burning around the coastal rim of eastern Australia, and the smoke covering the entire region shown in the map. Also shown in Figure 1 are the PM_2.5 particulate matter (particles with a diameter of ≤2.5 μm) concentrations over the region, predicted using the Australian Smoke Forecasting System (AQFx) that compares well to the smoke distribution shown on the Himawari-8 image. Information from AQFx was incorporated into air quality warnings issued by some State regulatory authorities during the bushfire event.

AQFx and the 2019/2020 summer fires

Figure 1. Himawari-8 visible image (red-green-blue composite imagery) of 21 December 2019. The AQFx forecast is an overlay of the national 27-km domain, the eastern 9-km domain, the Vic-Tas 3-km domain and the New South Wales 3-km domain.

Data from one air-quality monitoring station (AQMS), operated by regulatory authorities in each of the states of Queensland, New South Wales, Australian Capital Territory and Victoria are also presented here to illustrate the extent of the impact on air quality on south-eastern Australia.

The time series of PM_2.5 particulate matter measured at each AQMS shown in Figure 2 illustrates the progression of the impact of smoke on air quality from the north to the south of south-eastern Australia, from early in the fire season in November 2019, when PM_2.5 concentrations peaked in Queensland, to peaks in New South Wales in December 2019, to extremely high concentrations in the Australian Capital Territory in early January 2020 and peaks in Victoria in early to mid-January 2020. By the end of January 2020, PM_2.5 concentrations had returned to more typical values.

Figure 2. Time series of daily averaged PM2.5 concentrations measured at AQMSs in Queensland, New South Wales, the Australian Capital Territory and Victoria. The following services are acknowledged for their diligent efforts in collecting these data sets: Queensland Department of Environment Land and Water, New South Wales Department of Planning Industry and Environment, Australian Capital Territory Health Protection Service and Australian Capital Territory Environment Protection Authority.

The maximum daily PM_2.5 concentrations during the smoke plumes measured by the AQMSs in all states were very high, in Queensland and New South Wales four times greater than the national standard (that is, a national environment protection measure for PM_2.5 of 25 µg m^-3 24-hour average), in Victoria eight times greater than the national standard, and in the Australian Capital Territory almost forty times greater. At the AQMS in the Australian Capital Territory the daily average PM_2.5 concentration exceeded the national standard for 53 days between 1 November 2019 and 28 February 2020.

Smoke also had an impact on the Cape Grim Baseline Air Pollution Monitoring Station located in north-west Tasmania. Cape Grim is one of the Global Atmosphere Watch (GAW) Global Stations and has been operating for forty-four years. The hourly averaged black carbon (BC) time series for the period December 2019–February 2020 (Figure 3) shows smoke reaching Cape Grim on four occasions, with the highest BC concentrations observed on 3 January, peaking at over 4 µg m^-3 (hourly averaged). The longest duration of smoke reaching the station occurred between 6–10 January (although with low BC concentrations) and 13–16 January (with higher BC concentrations). High BC concentrations were also observed on 31 January. Also shown in Figure 3 are air history maps for one hour in each of these periods, which show the circulation of air over south-east Australia before reaching Tasmania. The time series of BC is coloured by mean hourly ozone (O₃) concentrations and shows an increase in ozone concentration measured at Cape Grim over the course of January. On 15, 16 and 31 January the ozone mole fraction levels reached 60 ppb or more. While these levels have been seen before during the four decades of ozone monitoring at Cape Grim, they have been exceptional events generally years apart. In the present case, they are associated with ageing of the smoke reaching Cape Grim.

Figure 3. Time series of BC plotted as a function of ozone concentration (presented by the colour scale) at the Cape Grim Baseline Monitoring Station during January 2020. Air history maps from the UK Met Office. Source: Numerical Atmospheric-dispersion Modelling (NAME), UK Met Office, courtesy of A. Manning.

More than 10 million people are likely to have experienced some exposure to these very hazardous concentrations of PM_2.5. Arriagada et al. (2020) have recently estimated that this exposure may be responsible for approximately 400  excess deaths, 1  120 hospital admissions for cardiovascular problems, 2 030 admissions for respiratory problems, and 1 300 emergency department attendances for asthma.

Figure 4. Daily fire emissions for August– October 2015 for carbon monoxide and biomass-burning aerosols. Exceedances of the 2003–2014 average are in red. 1 August 2015 Source: Adapted from Benedetti et al., 2016

In 2015, extensive burning of peat occurred throughout large parts of Indonesia from August to November. Peat fires are extremely difficult to extinguish and can burn continuously until the return of the monsoon rains. The strength and prevalence of these fires are strongly influenced by large-scale climate patterns such as El Niño (Field et al., 2004; van der Werf, 2008). Despite this inherent predictability, the 2015 fires in Indonesia escalated to an environmental and public health catastrophe (Field et al., 2016). Fire emissions during that period were consistently and extraordinarily strong, as shown by the number of days in 2015 when daily emissions of carbon monoxide (CO) and biomassburning aerosols (BC and organic matter (OM)) exceeded the maximum daily emissions during the same days in 2003–2014, as estimated by the Global Fire Assimilation System (GFAS) (Figure 4).

Total column carbon monoxide anomalies that reached 500% in the core of the fire region were remarkable, but even more striking were the extremely large anomalies (~2 000%) in total aerosol optical depth at 550 nm for biomass-burning (OM + BC) aerosols that covered large areas of the Indian and western Pacific Oceans (Benedetti et al., 2016).

In addition to the air quality and climate implications of such events, which add large amounts of aerosols to the atmosphere, the weather is also altered. During the 2015 Indonesia peat fires, surface cooling related to smoke from the fires was evident. For accurate weather predictions during such events, aerosols need to be considered in the forecast, as demonstrated by studies made by the European Centre for Medium-Range Weather Forecasts (ECMWF). Experiments to investigate the impact of this large aerosol load on sub-seasonal-to-seasonal prediction were run using the ECMWF ensemble prediction system (Benedetti and Vitart, 2018). Re-forecasts for October 2015 conditions using observed emission from GFAS clearly show the pattern of surface cooling associated with the fires three months ahead (Figure 5). An aerosol climatology is not useful for these extreme cases, and a fully integrated fire dynamical model to predict emissions along with radiatively interactive prognostic aerosols would be needed for the prediction to be accurate.

Figure 5. Observed fire radiative power (top) from MODIS imaging for the period August–October 2015 and (bottom) surface temperature anomalies forecasted for October 2015 three months in advance . The surface cooling over Indonesia corresponding to the smoke from the fires is clearly evident.

Figure 6. Total emissions of particulate matter for June–August from wildfires within the Arctic Circle between 2003 and 2019. Emissions are estimated with the CAMS Global Fire Assimilation System v1.2, which uses MODIS satellite observations of fire radiative power.

Large fires in the boreal forests of Eurasia and North America are a common occurrence from late spring through to early autumn. These fires can release large quantities of smoke to the atmosphere where they may be subject to long-range, even intercontinental, transport over thousands of kilometres. Recent summers have seen significant fires in boreal forests in North America and Siberia and poleward of the Arctic Circle (total fire emissions are shown in Figure 6), smoke from which has been observed to be transported high into the Arctic Circle and occasionally to undergo transpolar transport. Increasing boreal forest fire activity and transport of smoke into the Arctic Circle have potential climate impacts through increased surface deposition of BC and particulate matter on to sea ice, affecting the albedo, that is, the amount of solar radiation reflected, leading to enhanced warming and melting, and changes in absorption/reflection of atmospheric radiation with potential impacts on polar meteorology and climate.

Estimates of emissions are available from satellite observations of either the burnt area or fire radiative power (a measure of the rate of emitted radiative energy by the fire at the time of the observation). These observations can be used to estimate the amount of vegetation consumed by fires, information that can be further used to estimate the amount of carbon, gases and aerosols released to the atmosphere. A number of systems for estimating fire emissions based on these observations exist although for real-time applications this is only feasible with fire radiative power observations. The requirement for real-time fire emissions estimates arises from the development of real-time forecasting systems for atmospheric composition and air quality. The European Commission’s Copernicus Atmosphere Monitoring Service (CAMS), implemented by ECMWF, is one such service and makes use of a wide range of satellite observations to produce analyses and 5-day forecasts. Forecasts from CAMS provide information on BC and organic matter aerosols and allow the monitoring of smoke transport from fires worldwide.

The intensity of solar radiation passing through the atmosphere is reduced by aerosols via absorption and scattering. The combined effect, known as radiation extinction, depends on the amount and absorption properties of the aerosols present.

Figure 7. Mean tropospheric mass mixing ratio of organic matter aerosol (in units of 10E-9 kg/kg) between 6–12 July 2015 based on CAMS analyses and showing the extent of the smoke plume from Alaska to the European Arctic.

The importance of smoke aerosol for an accurate radiation forecast is demonstrated in a case study of trans-polar smoke transport from Alaska to the European Arctic via the eastern Arctic Ocean in July 2015 (Figure 7). The impact of this event on the forecast is a significant increase in short-wave radiation extinction (with as consequence a decrease in Earth warming) with an average decrease in the surface solar radiation of 5–10 W/m² local to the plume during its transport.

The impacts that smoke aerosols have on radiation in the high Arctic were studied using model experiments performed at ECMWF. To this end, a model simulation of the specific smoke event has been compared to a climatological simulation.¹ The simulations employing different aerosol fields reveal how radiation in the model reacts to that specific change, thereby offering the possibility to quantify its effect. In particular, the model’s surface short-wave radiation response to different levels of aerosols in the atmosphere was analysed. To analyse a specific event, the prognostic aerosol fields² of that event, available from CAMS, were compared to climatological ones.¹

The plume also reduced the region’s overall radiation reflected at the top of the atmosphere by about 10–20  W/m² and increased absorption within the aerosol layer (Figure 8). July 2015 was the second worst fire year for the United States of America's state of Alaska, following 2004, but the equivalent experiments for Arctic smoke plumes in other years have yielded similar results.

Figure 8. The impact of smoke aerosol on radiation forecasts. Shading shows the difference between a 24-hour forecast with prognostic aerosol in the radiation scheme and another using a monthly mean aerosol distribution for (a) surface downward shortwave radiation and (b) top-ofatmosphere net short-wave radiation, averaged from 8 to 13 July 2015. Contours show organic matter aerosol optical thickness at 550 nm for values of 0.1, 0.2, 0.4, 0.8 and 1.0. Net radiation is defined as downwelling minus up-welling fluxes.

____________________________

¹ Long-term average.

² Prognostic fields are fields generated by a model simulation, such as from a forecast model.

We currently live in a golden age of Earth observations, with a global observation system that has grown and been greatly expanded over the last 20 years, including the availability of active fire data exploitation of the wide range of currently available Earth fire observations that provides essential information on global fire activity and emissions. The available datasets of global fire emissions have typically provided daily estimates based on observations made from low-Earth-orbit geosynchronous satellites,¹ which produce information at the same local time each day. The CAMS Global Fire Assimilation System has recently been redeveloped to estimate fire emissions at an hourly time resolution using a parametrized² diurnal cycle and also including the capability to exploit additional low-Earth-orbit observations with improved sensitivity as well as geostationary satellite observations with an increased time resolution of approximately 10 minutes. Combining data from different satellite sensors in this way to provide increased timeliness of global fire data will provide valuable information for civil protection and air-quality monitoring, particularly with the planned launch of air quality observations from geostationary satellites. Further improvements the monitoring of biomass-burning aerosols are also expected with improved information on emission factors from different vegetation types; recent laboratory and field studies have shown that not taking vegetation differences into account has led to an underestimation of particulate matter emissions in some regions.

In response to the Indonesian wildfire crisis of 2015, WMO has initiated a Vegetation Fire and Smoke Pollution Warning and Advisory System (GAW report 235, see Goldammer et al., 2018) with the aim of providing guidance to affected WMO Member States through specialized regional centres. The first Regional Vegetation Fire and Smoke Pollution Warning and Advisory Centre has been established for the WMO South-West Pacific Region V, operated by the Meteorological Service of Singapore. Forecasts and timely information about wildfires are provided as part of the service. The CAMS smoke forecasts and GFAS emissions are also made available to the Singapore Met Service. Future collaborations will involve providing data to the other regional centres currently being established.

___________________

¹ A geosynchronous satellite is a satellite in geosynchronous orbit, with an orbital period equal to the Earth's rotation period. Such satellites return to the same position in the sky after each sidereal day.

² Parametrization in a weather or climate model is a method of replacing processes that are too small scale or complex to be physically represented in the model by a simplified process.

Arriagada B.A., A.J. Palmer, D. Bowman, G.G. Morgan, B.B. Jalaludin and F.H. Johnston, 2020: Unprecedented smoke‐related health burden associated with the 2019–20 bushfires in eastern Australia. Med. J. Aust, 213(6):282– 283

Benedetti, A., F. Di Giuseppe, J. Flemming, A. Inness, M. Parrington, S. Rèmy and J.R. Ziemke, 2016: Atmospheric composition changes due to the extreme 2015 Indonesian fire season triggered by El Niño. In: State of the Climate in 2015 (J. Blunden and D.S. Arndt, eds.). Bull. Amer. Meteor. Soc., 97(8):S56–S57

Benedetti A. and F. Vitart, 2018: Can the direct effect of aerosols improve subseasonal predictability? Mon. Wea. Rev., 146(10):3481–3498

Field, R.D., Y. Wang and O. Roswintiarti, 2004: A droughtbased predictor of recent haze events in western Indonesia. Atmospheric Environment, 38(13):1869–1878.

Field, R.D. et al., 2016: Indonesian fire activity and smoke pollution show persistent non-linear sensitivity to El Niñoinduced drought. PNAS, 113(33):9204–9209

Goldammer, J.G. et al., 2018: Vegetation Fire and Smoke Pollution Warning and Advisory System (VFSP-WAS): Concept Note and Expert Recommendations. GAW Report No. 235. Geneva, WMO

van der Werf, G.R., et al., 2008: Climate regulation of fire emissions and deforestation in equatorial Asia. PNAS, 105(51):20350–20355

This Bulletin was written by Angela Benedetti (ECMWF), Mark Parrington (ECMWF), Francesca Di Giuseppe (ECMWF), Melita Keywood (Commonwealth Scientific and Industrial Research Organisation) and Olga L. MayolBracero (University of Puerto Rico) with support of the GAW Scientific Advisory Group on Aerosols.