Tropical African wildfire aerosols trigger teleconnections over mid-to-high latitudes of Northern Hemisphere in January

This study investigates the impacts of African wildfire aerosols (primary organic carbon, black carbon and sulfate) on the Northern Hemispheric in January. We found that wildfire aerosols emitted from equatorial Africa result in two mid-to-high latitudes atmospheric Rossby wave trains. One is from subtropical Atlantic propagating northeastward across Europe to Siberia, and the other one propagates eastward from Middle East across Asia to Pacific Northwest. The maximum positive geopotential height anomaly locates in Europe, concurrent with a greater-than-2 K land surface warming. These Rossby wave trains are excited by the atmospheric heating that caused by the wildfire aerosols in equatorial Africa and propagate into extratropics with the help of the westerly jet. Based on the diabatic heat budget analysis, the Rossby wave source is primarily from the solar absorption of black carbon of African wildfire. The present study emphasizes that wildfire aerosols, especial the absorbing aerosols, would have profound climate effects on remote regions and thus need more attentions.


Introduction
Aerosol through scattering and absorbing solar radiation could alter energy balance, and also can be activated as cloud droplet nuclei or ice nuclei, changing cloud microphysical properties and affecting cloud radiative effects. Although the aerosol radiative effects and aerosol-cloud interactions have been widely studied and evaluated, they are still long-lasting hot topics in the atmospheric community. The radiative forcing of aerosols could induce regional warming or cooling (e.g. Ramanathan and Carmichael 2008, Chen et al 2016, and the persistence of the regional anomalies could further induce large-scale climate response via dynamic or thermodynamic processes.
Great efforts have been devoted to investigating the regulation of aerosol effects on regional climate. For example, the deposition of dust and black carbon (BC) on Tibetan Plateau could amplify the heat pump effect, which would advance the onset of South Asian monsoon, and affect East Asia monsoon . BC emitted over India could advance the Indian monsoon onset (Bollasina et al 2013), leading to increased (suppressed) pre-monsoon (monsoon) rainfall (Ganguly et al 2012) through absorbing solar radiation and increasing the meridional temperature gradient (Meehl et al 2008). The impacts of anthropogenic aerosols on East Asian monsoon has been also studied widely (Qian et al 2001, Giorgi et al 2002, Liu et al 2009, Zhang et al 2012, Jiang et al 2013, Song et al 2014. A more recent atmosphere-ocean coupled modeling study (Lou et al 2019) showed that BC heating over North China can weaken East Asian winter monsoon through changing land-sea thermal contrast and cloud feedbacks. In addition, polluted snow with light-absorbing aerosols over Tibetan Plateau could decrease snow albedo, leading to reduced snow cover that in turn changes the East Asian climates (Qian et al 2011, Yasunari et al 2015. Aerosols not only exert local climate impacts on surrounding regions , Zhou et al 2020, but also generate global impacts, such as weakening the Hadley Cell Sherwood 2011, Tosca et al 2013), and shifting the subtropical westerly jets (Ming et al 2011). A few works have reported that teleconnections may be induced by aerosols. Rodwell and Jung (2008) suggested the role of Sahara dust in driving the planetary waves. Kim et al (2006) reported that the direct radiative effects of dust and BC could excite a planetary wave spanning from North Africa through Eurasia to the North Pacific in boreal spring. However, these two studies only considering the absorbing aerosols. What are the global circulation responses when scattering aerosols and the indirect radiative effects of aerosols are also involved? This is an interesting issue merits further investigation.
This work is a follow-up study of Jiang et al (2016) in which the direct and indirect effects of three types of fire aerosols (sulfate, BC, and primary organic matter) are investigated. The indirect effects of fire aerosol are found to be much larger than its direct effects. With the basic understanding of the thermodynamic properties of Africa wildfire aerosols, in the present study we aim to unravel how the remote climate responses (the teleconnections) of aerosols are established via numerical simulations. The reasons for choosing January as the target study period are two folds. First, the African wildfire mostly occur in boreal winter with the peak in January. Second, given that North Africa is the only major wildfire source on the earth during January, it is a time window to identify the remote climate impacts of biomass burning aerosols. The remainder of this paper is organized as follows. Section 2 introduces the overview of model and experiments. Section 3 presents the radiative effects of the Africa wildfire aerosols and corresponding circulation changes, and explores the cause of the atmospheric teleconnections. The conclusion and discussion are provided in section 4.

Model
The Community Atmosphere Model version 5.3 (CAM5.3) (Neale et al 2010) with the finite volume dynamics core are used. A two-moment cloud microphysics scheme (Morrison et al 2008) is adopted in the model. The Modal Aerosol Model (MAM4) that consists of four lognormal modes (Aitken, accumulation, coarse, and primary carbon mode) is used to predict aerosol mass and number mixing ratios . The primary carbon mode is included to improve the treatment of microphysical ageing of BC and POM .

Results
Boreal winter is the fire season over North Africa (van der Werf et al 2017), and the wildfire mostly occur in January with its carbon emissions up to 160 Tg C per month (see the supplementary material figure S1, which is available online at stacks.iop.org/ERL/16/034025/mmedia). At that time, Africa is the major fire hotspot ( figure 1(a)).
Considering the lifetime of wildfire aerosols in troposphere is from several days to 2 weeks, the corresponding atmospheric responses are also within the timescale of one month (Jacob 1999). Thus, in the following paper, we only show differences in January between Fire and noFire experiments to illustrate the African wildfire climate impacts. As the aerosols have significant impacts on the radiative budgets, we present the radiative effects as well as the direct and indirect effects in section 3.1, demonstrate the impacts on atmospheric circulation in section 3.2,  and link the aerosols with the circulation anomalies in section 3.3. Figure 1 shows the impact of fire aerosols on the aerosol optical depth (AOD), radiative fluxes, and surface temperature. The maximum changes of AOD (dAOD > 0.25) is located in the equatorial Africa and the downstream region of the entire tropical Atlantic ( figure 1(a)). The fire aerosol effects on the net radiative flux (FNET) at the top of the atmosphere (dFNET_toa) and at the surface (dFNET_sfc) are shown in figures 1(b) and (c), respectively. The maximum negative dFNET_toa (up to −10 W m −2 ) is consistent with the maximum increase of AOD over tropical Atlantic. The second maximum negative dFNET_toa (around −2 W m −2 ) appears over Europe, where the increase of AOD is weak. The dFNET_sfc can be −20 W m −2 over the tropical Atlantic. The pronounced difference between dFNET_sfc and dFNET_toa over the equatorial Atlantic implies that the fire aerosol absorption in the atmosphere is quite strong. Significant changes in surface air temperature (figure 1(d)) can be found over mid-to-high latitudes of Europe, Northern Russian, and Alaska. However, the changes in AOD and FNET_sfc are not significant in these regions, and the negative change in FNET_toa will result in an overall cooling. This suggests that the local changes in radiative fluxes cannot explain the changes in surface air temperature in the regions, especially for the remarkable warming over Europe. We decompose the total aerosol effects at the top of atmosphere (TOA, dFNET_toa) into direct and indirect effects, and surface albedo effects following the method of Ghan (2013) to understand the direct and indirect effects of aerosols. The total effects (sum of shortwave (SW) and longwave (LW) radiative effects) are shown in figures 2(a)-(d). The indirect effects of aerosols (figure 2(a)), caused by aerosol-cloud interactions and their impact on radiation, are most profound at the downstream Atlantic Ocean with the maximum of up to −10 W m −2 . The direct effects of aerosols (figure 2(b)), which are mainly due to the direct absorption and scattering of radiation by aerosols, are mostly confined to the regions where AOD are significantly increased. The contrast between all-sky (positive, figure 2(b)) and clear-sky (negative, figure 2(c)) direct effects in equatorial Atlantic indicates that clouds enhance the absorption by aerosols above the cloud top (Lu et al 2018) and it may be responsible to the large uncertainties in evaluating the total effects of aerosols (Boucher et al 2013). The effects of aerosols on surface albedo is shown in figure 2(d), which show profound responses in mid-to-high latitude, such as Europe and North America (dominated by the LW component, figure 2(f)). The significant change in LW surface albedo effects in Europe (figure 2(f)) is consistent with the surface warming shown in figure 1(d), suggesting that the warming surface emit much more LW radiation flux and enhance the radiative cooling effects at the TOA.

Fire aerosol-induced radiative fluxes
The above results indicate that the remote responses in total TOA radiative forcing are mainly from the surface albedo effects and surface temperature changes. However, the local radiative effects cannot in turn explain the significant remote surface temperature warming over Europe. Therefore, in the next subsection, we will explore how these remote responses of surface temperature anomalies are generated.

Fire aerosol-induced teleconnection patterns
As discussed in section 1, despite of the radiative effects, aerosols could have significant impacts on atmospheric circulations due to the nonhomogeneous spatial distribution radiative forcing. Thus, the circulation responses to African wildfire aerosols are analyzed in this section.
The zonal wind from the noFire experiment and the geopotential height anomalies at 300 hPa due to fire aerosols (Fire experiment minus noFire experiment) are presented in figure 3(a). Significant atmospheric responses to the wildfire appear over the and geopotential height anomalies at 300 hPa due to fire aerosols (Fire experiment minus noFire experiment) (contour, at the internal of 3 m), red contours indicate the changes of aerosol optical depth (dAOD, only greater than 0.05 are shown); (b) the vertical cross section of geopotential height (color shading) along the centers (A to E) of the geopotential height anomalies in (a); (c) the spatial distribution of wave activity flux (vector, units: m 2 s −2 ) and its divergence (color shading, units: 10 −7 m s −2 ). The geopotential height differences passing the 95% confidence level in (a) and (b) are dotted, and the wave activity flux and its divergence between 5 • S and 5 • N in (c) are masked out. mid-to-high latitudes of Northern Hemisphere with large magnitudes in North Atlantic, Europe, and Gulf of Alaska and the Pacific Northwest. The circulation changes are in well-organized wave-train patterns, in which the anomaly from North Atlantic across Europe to Siberia resembles the Eurasia teleconnection pattern (Liu et al 2014, Wang andZhang 2015), and the anomaly over North America is more like a Pacific-North America (PNA) pattern (Wallace andGutzler 1981, Leathers et al 1991) which ends at the southeast United States.
With the zonal wind at 300 hPa, we could further understand how such regional circulation changes are formed. The geopotential heights with small magnitudes from the Middle East to East Asia corresponds with the westerly jet stream over Eurasia. At the exits of the Atlantic westerly jet and Asian westerly jet where locate the large horizontal gradients of the basic mean flow, the magnitude of the perturbation is amplified, resulting in significant geopotential height anomalies over North Atlantic, North Eurasia. These changes of circulations in the mid-to-high latitudes are barotropic in vertical ( figure 3(b)). The more than 2 K land surface warming over Europe locates between the cyclonic and anticyclonic anomalies of the circulation changes (figures 1(d) and 3(a)). Then, a nature question arised as to how does the atmospheric teleconnection induce the regional change in the surface air temperature, especially for the 2 K land surface warming over Europe? To answer this question, we diagnosed the relative contributions of each term in temperature equation: the temperature advection, the adiabatic heating, and the diabatic heating (see the supplementary material figure S2). The result shows that the advection term is the main contributor to the land surface warming over Europe, while the diabatic heating and adiabatic heating terms are much smaller than the advection term by two orders of magnitude. Therefore, the temperature advection that induced by the barotropical atmospheric teleconnections is mainly responsible for the remarkable Europe surface warming.
The wave activity flux defined in Takaya and Nakamura (2001) are calculated to identify the wave source and the propagation pathways of the Rossby waves. As shown in figure 3(c), strong wave activity fluxes are observed from the subtropical Atlantic to Europe, and from the gulf of Alaska to the Pacific Northwest, corresponding to the regions with robust circulation anomalies ( figure 3(a)). For the wave train from the Middle East to East Asia, wave activity fluxes are relatively weak. Generally, there are two Rossby wave trains (figure 3(a)). One propagates northeastward to the northeast Atlantic, and further eastward to North Europe and Siberia. The other is a minor branch which propagates eastward, across Middle East to East Asia, and then further northward, bending equatorward over along the west coast of North America. For the minor branch, it can be found that the Asian westerly jet act as the waveguide for the wave propagation, which is consistent with previous studies (Lu et al 2002, Enomoto et al 2003. Two major divergent regions of wave activity flux can be found from figure 3(c), one is over subtropical North Atlantic, and the other is over North Pacific. Given that the circulation changes are all derived from the effects of the African wildfire, we infer that the subtropical North Atlantic (60 • -30 • W, 20 • -40 • N) located to the northwest of the wildfire aerosol source is the Rossby wave source region, and the anticyclone anomaly (figure 3(a)) is a direct tropical descending Rossby wave response to the atmospheric heating (Gill 1980) induced by the Africa wildfire aerosols.

Trigger of the teleconnections
From the wave activity flux analysis, we identified the Rossby wave source stems from the anticyclonic anomaly over the subtropical North Atlantic. Based on the Gill-type mechanism (Gill 1980), the anticyclonic anomaly is induced by the atmospheric heating. Thus, we first analyse the diabatic heating budget over the African aerosol source region, and then try to understand how the aerosols perturb the atmosphere and trigger the teleconnections in the mid-tohigh latitudes. Figure 4(a) illustrates the vertical profiles of air temperature (T), solar heating rate (QRS), longwave heating rate (QRL), moist process heating rate (DTCOND), and the temperature tendency from the diffusion (DTV) over equatorial African wildfire regions. A peak warming of 0.2 K in T appears at 700 hPa. QRS shows significant increase below 600 hPa, while QRL has a negative peak at 850 hPa, and DTV is strongly negative near surface due to strong radiative forcing at surface. The DTCOND is negative below 500 hPa. In sum, for the atmospheric heating over the equatorial African wildfire region, QRS is the dominant contributor with its maximum at lower level troposphere.
As a strong absorbing aerosol, BC could enhance the absorption of solar radiation. It can be found that the enhanced QRS is mainly caused by the effect of BC ( figure 4(b)). The magnitude of BC-induced solar heating rate can reach to 0.16 K d −1 , accounting for about 80% to the total aerosol effects on QRS. Thus, we conclude that the absorption of wildfire aerosols by BC mainly contributes the atmospheric heating over equatorial African wildfire region. In fact, Tosca et al (2013) and Jiang et al (2016) have also found the surface temperature warming over Europe as the response to direct and semi-direct effects of the aerosols (although the signals are less robust based on annual mean results). Such similarity provided us confidence on that the direct effect of wildfire aerosols (i.e. BC absorbing) induces the circulation anomalies and 2 K land surface warming over Europe in January.
To further validate that the teleconnections are triggered by the atmospheric heating induced by African wildfire aerosols. An intermediate AGCM was employed to study the formation mechanism of the teleconnection pattern. This model was based on the spectrum dry AGCM (Held and Suarez 1994) with five evenly distributed sigma levels from 0.9 to 0.1 and a horizontal resolution of around 2.88 • × 2.88 • . The climatological mean state is taken from that of the ensemble mean of the noFire experiments in January. A damping rate of 1 d −1 in the lowest level (σ = 0.9) and linearly decaying to 0.1 d −1 at the middle level (σ = 0.7) are taken as Rayleigh friction to apply to the momentum equations and to mimic the planetary boundary layer. Newtonian cooling with an efolding time scale of 10 d is applied to the temperature equation at all model levels. The heating has a cosinesquared profile in an elliptical region over equatorial Africa in horizontal and the vertical profiles are given as in figure 4(c) with the maximum rate of 2 K d −1 at the level of σ = 0.7 to minic the atmospheric heating induced by the wildfire aerosols.
The time evolution of 300 hPa geopotential height response to the specified heating over the equatorial Africa are shown in figure 5. As expected, a northwestward-propagating Gill-type tropical Rossby wave duplet is induced over both sides of the equator to the west of the heating center. Meanwhile, a tropical Kelvin wave response is generated at the east of the heating center, propagating eastward. When the Rossby wave arrives around 30 • N, it perturbs the westerly jet stream (figures 3(a) and 5(d)).
The westerly jet stream over North Atlantic leads to a northeastward propagation of the Rossby wave, forming the northern branch of the Rossby wave trains. The westerly jet stream over Eurasia acts as the waveguide and transports the perturbation to the East Asia, Pacific, and further to the west of the North America. The teleconnection patterns are quite steady after around 20 days integration. Compared to figure 3, the geopotential height anomalies over Europe, Pacific and North America from the idealized model are relatively weak. These discrepancies are possibly due to the southeastward shift of the anticyclone over the subtropical North Atlantic in the model, which limits the energy being transported northeastward by the westerly jet. Despite of such discrepancies, this idealized model simulation confirms that the atmospheric heating over equatorial Africa (induced by the wildfire aerosols) could generate the upper level anticyclone anomaly over the subtropical North Atlantic and further trigger the two Rossby wave trains in the mid-to-high latitudes of the Northern Hemisphere.

Conclusion and discussion
Using the global climate model CAM5, we evaluated the remote effects of wildfire aerosols emitted from equatorial Africa in January. Wildfire aerosols cause significantly radiative forcing over the source region and the downwind region over Atlantic, and further induce remote effects in the mid-to-high latitudes with a greater-than-2 K land surface warming in Europe. The remote response is imposed through atmospheric teleconnections in terms of Rossby wave trains. Through analyzing the wave activity flux and idealized AGCM simulation, we demonstrated that the equatorial African aerosol-induced low-level troposphere atmospheric heating could result in an anticyclonic anomaly over the subtropical North Atlantic. The anticyclonic anomaly perturbs westerly jet, and excites two Rossby wave trains. One Rossby wave train is along the great circle route via the northeast Atlantic and Europe to Siberia and the North Pacific, while the other is along the Asia jet stream eastward and northeastward with a shorter wave length toward the North Pacific, and then bending equatorward to Pacific Northwest. The budget analysis of the diabatic heating over the African aerosol source region indicates that the absorption of solar radiation by BC can contribute about 80% of the total changes of solar heating rate which plays the key role in stimulating the teleconnections.
Based on the Rossby wave theory, Rossby waves hardly escape from the tropics to the extratropics with a tropical easterly mean flow. Two possible mechanisms may explain our findings. One is that the vertical wind shear of the tropical mean flow can couple the baroclinic and barotropic Rossby wave modes; the baroclinic mode is trapped near the equator but the barotropic mode can emanate from the tropics to extratropics . The other one is that a southerly component in the basic easterly flow (a southerly conveyor) may transfer a Rossby wave source northward. Thus, a forcing embedded in the deep tropical easterlies may also excite a Rossby wave in the extratropical westerlies (Wang et al 2005).
Note that the strong negative radiative forcing over the equatorial Atlantic may further induce low frequency variability. Booth et al (2012) suggested that the anthropogenic aerosols may contribute to the multi-decadal variability of the North Atlantic SST. Amiri Farahani et al (2020) found the southern Africa fire aerosols could result in a La Nina-like SST response in a ocean coupled model. In this study, as we prescribed global SST, the strong cooling effects over the equatorial Atlantic are suppressed, and thus the Rossby wave trains we discussed in the present study are a fast response of the atmosphere. When an ocean model is coupled with the atmospheric model, the strong cooling effects on the Atlantic SST may induce low frequency variability which would combine the fast response of the atmosphere. We would not go any further for the coupled model in this study, but this may merit further investigations.
The waveguide effect of westerly jet in conveying the influences of tropical SST-forced diabatic heating to the globally climate variations (Li et al 2019) has been extensively investigated. For example, The Atlantic SST anomalies could excite eastwardpropagated Rossby wave trains across the Eurasian continent throughout the year , and the Rossby wave trains could be found on multiple timescales (e.g. from intraseasonal to interannual, and from decadal to multidecadal). In boreal summer, the Eurasian teleconnection pattern is associated with Atlantic SST (Lu et al 2002, Enomoto et al 2003, Sun et al 2015, while in boreal winter, an Africa-Asia teleconnection pattern embedded in the Asian jet is formed with the forcing of Atlantic SSTA (Sun et al 2017). The anomalous upper-level divergence/convergence associated with the SST warming/cooling contributes to Rossby wave source, whilst the westerly jet acts as a waveguide for the propagation of the Rossby wave, shaping the pathway of the teleconnections (Zhu and Li 2016). In the present study, we emphasis that besides the SST forcing, the heating effect of the African wildfire aerosols can also trigger these teleconnections with the help of the westerly jet in January.
Another point we would like to emphasize here is that although the total radiative forcing of wildfire aerosols (figure 2) (Jiang et al 2016) is cooling, what triggers the planetary Rossby waves and remote effects is the anomalous atmospheric heating induced by the absorbing aerosols. Thus, the existence of the absorbing aerosols may need more cautions and attentions. Furthermore, the 2 K land surface warming over Europe would further enhance snow melting and reduce the snow cover and duration, leading to consequent ecological and socioeconomic impacts regionally and globally (Barnett et al 1988, Bamzai andShukla 1999).

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: www.globalfiredata.org.