Absorption of shortwave solar radiation can potentially heat aerosol layers and create buoyancy that can result in the ascent of the aerosol layer over several kilometres altitude within 24–48 hours. Such heating is seasonally dependent with the summer pole region producing the largest lifting in solstice because aerosol layers are exposed to sunshine for close to 24 hours a day. The smaller the Angstrøm parameter, the larger the lifting potential. An important enhancement to lifting is the diffuse illumination of the base of the aerosol layer when it is located above highly reflective cloud layers. It is estimated that aerosol layers residing in the boundary layer with optical properties typical for biomass burning aerosols can reach the extra tropical tropopause within 3–4 day entirely due to diabatic heating as a result of solar shortwave absorption and cross-latitudinal transport. It is hypothesized that this mechanism can explain the presence and persistence of upper tropospheric/lower stratospheric aerosol layers.
 Seasonal injection of biomass burning aerosol into the atmosphere occurs in many regions of (sub)tropical vegetation. Biomass burning aerosols contain large quantities of black carbon and organic carbon. Their optical properties are different from inorganic aerosols because they have a comparatively low single scattering albedo and are mostly hydrophobic. This means that, contrary to their inorganic counterpart, they absorb solar radiation, and will be resistant to serve as cloud nuclei. It is common knowledge that aerosols alter the reflective properties of the atmospheric column, so that the radiative balance is changed. Less attention has been given to the question of what happens to the aerosol layer itself when it is heated. Radiative heating is a diabatic process that alters the buoyancy of the layer resulting in a potential vertical displacement. So, aerosol layers, once they drift away from their locus of origin, potentially ascend/descend towards different levels of the troposphere. If the time scale of vertical displacement is shorter than the residence time of aerosols in the atmosphere, this is a powerful mechanism to distribute aerosol layers throughout the depth of the troposphere.
 This concept was first discussed in the Nuclear Winter literature in the 1980's [Malone et al., 1986; Kao et al., 1990], but as Nuclear Winters failed to materialize it was never verified by atmospheric data. Evidence of the lifting of aerosol layers by solar heating in real atmospheric conditions was presented by Radke et al.  and Herring and Hobbs  in relation to the burning of aviation fuel and the Kuwait oil fires.
 Although there is now enough evidence of injection of biomass burning aerosol into the lower stratosphere [Fromm et al., 2000] it has been suggested that such excursions are the result of pyro - convection [Fromm et al., 2005]. While this is a plausible mechanism for deep vertical transport of aerosols, it is unlikely that every biomass burning plume can instantaneously overcome the typical 60 to 80 K buoyancy jump between the surface and stratosphere. Deep convection is patchy, and dependent upon the thermodynamic stability of the atmosphere. Conditions favourable for biomass burning events (dry, hot, high pressure systems) are often unfavourable for the onset of deep convection so that the injected aerosol layers cannot reach altitudes higher than 4–5 km, i.e., the top of the dry boundary layer under hot conditions. So, additional alternative physical mechanisms should be put forward to explain stratospheric aerosol layers.
 The purpose of this paper is to demonstrate that it is feasible for aerosol layers to reach the upper regions of the troposphere and the lower stratosphere due to solar heating alone, thus without invoking concepts such as deep convection or synoptic cell circulations. It will be shown that layers initially present in the low troposphere can reach the top of the troposphere in less than four days which is shorter than a typical aerosol residence time. Section 2 outlines our method which is based on a combination of radiative transfer and thermodynamic theory. In Section 3 we will describe the results, including the sensitivity of solar absorption events to water vapor, cloud albedo. Results are summarized in Section 4.
 We first calculate the absorption characteristics of single aerosol layers as a function of latitude assuming fixed values of the single scattering albedo (SSA = 0.75) and Angstrøm parameter [α = 1.50], where α is defined as (τ/τ500nm) = (λ/500 nm)−α, and τ is the optical thickness, λ is wavelength (in nm). Asymmetry parameter (g) was fixed at 0.7. Observations of the optical properties of biomass burning aerosols do not produce a unifying picture. Single scattering albedo for biomass burning particles/aerosols varies anywhere in the range from 0.6 to 0.9 [e.g., Johnson et al., 2008], although Mitchell et al.  put it in the upper part of this range, while α varies between 1 and 2. So, our ability to adequately calculate the directional characteristics of transmission, reflection and absorption functions is hampered. Therefore we employ a relatively simple delta-Eddington two-stream routine in 24 bands in the shortwave region of the solar spectrum [Boers and Mitchell, 1994] in favour of a more complex radiative transfer computation. The following idealized conditions were considered: a) The aerosol layer stays undiluted for several days, b) Rayleigh scattering, water vapor and ozone absorption, variations in surface albedo and longwave radiative transfer are neglected, c) SSA and g are spectrally constant, and d) The entire layer is presumed to experience the same heating rate everywhere. These idealized conditions are discussed below.
 Next, the absorption is related to temperature changes and subsequently to altitude changes. Here, diabatic heating is decoupled from the adiabatic cooling process using the First Principle of Thermodynamics which relates the change of potential temperature to the amount of heat that is stored in a layer:
[Iribarne and Godson, 1981], where q is the heat content, cp is the specific heat capacity of dry air, p is pressure, p0 is reference pressure (1000 hPa), θ is potential temperature and κ = Ra/cp, Ra is the gas constant of dry air. If the only diabatic process is absorption of shortwave radiation, then
where ρ is density of dry air, R is the shortwave radiative flux, t is time and z is altitude in the atmosphere. Longwave radiation is neglected in (2) because, in considering differences between aerosol laden air and its environment, it remains largely unchanged. The static stability of the atmosphere requires that the virtual potential temperature θv = θ (1 + 0.609 qv) increases with altitude, where qv is the water vapor specific humidity. Neglecting the specific humidity term in the brackets (2–3% error), potential altitude variations can then be calculated by assuming that the changes in potential temperature due to radiation heating translate into an altitude gain up to the level at which the new potential temperature equals that of its environment:
where z1 is the new altitude of the layer, z0 is the old altitude of the layer, Γ the lapse rate of potential temperature, and the subscript 1, and 0 for the potential temperature designate the begin and end time of the solar heating (e.g., the duration of solar heating per day). The NCEP – reanalysis results of the latitudinal 50 – year average of the temperature was used to calculate the lapse rate [Kistler et al., 2001].
3.1. Solar Absorption and Latitude
 The absorptivity of an aerosol filled layer of 3 km thickness was calculated as a function of latitude. The value of 3 km was chosen to reflect a typical depth of a smoke-filled boundary layer. Here we use the situation at solar solstice point 21 June to avoid potential latitudinal symmetry. The absorptivity was defined as:
Abs is the total absorbed radiation, S ↓ is the downwelling flux at the top of the atmosphere. An aerosol layer with a constant extinction profile and a time integration step of 10 minutes was used in the calculation. Solar absorption is small for the winter hemisphere (see Figure 1), but for the summer hemisphere most absorption of the available solar radiation takes place at high latitudes. For optical thicknesses over 2.0, in excess of 20% of available radiation is absorbed for the SSA = 0.75, α = 1.50 at any latitude over 40 degrees, so that potential solar absorption is not confined to the tropical and subtropical regions of the globe.
3.2. Altitude Changes
 The NCEP reanalysis results were used to calculate the potential temperature as a function of altitude for all latitudes, again using the date of 21 June. Equations (1)–(3) were applied to convert the absorptivity into a heating rate and then into an altitude change. The 24 hour altitude change was plotted (Figure 2), again for SSA = 0.75, α = 1.50. Also an initial relative humidity of 60% was used in the calculation. The dry altitude gain reflects the 24 hour altitude change assuming no cloud formation. This is obviously incorrect for layers that reach saturation but we chose to ignore this process because condensation invokes an unnecessary complexity to the configuration of the model. For a case with a high concentrations of biomass burning aerosols, their lack of hygroscopicity and the relative abundance of particles would result in an intractable activation process. Also, as soon as cloud droplets, or ice particles are formed, infrared radiative cooling and shortwave reflection of solar radiation on the deliquescent aerosols would become important. Clearly, all these factors may play a role but detract from the purpose of this paper, which is the investigation of the relevance of solar absorption for the tropospheric ascent of aerosol layers.
Figure 2 demonstrates that a 24 hour altitude gain of 1–1.5 km is achievable over a large range of optical depth values, and for a large portion of the globe. At the higher end of optical depth values, altitude gains in excess of 2 km are possible, so that for cases when an aerosol layer remains intact for 3 or more days, such a layer could reach tropopause levels.
3.3. Single Scattering Albedo and Angstrøm Parameter
Figure 3 shows the altitude change in km per 24 hours for an aerosol layer of optical depth 1.75, at a latitude of 40 degrees on 21 June. Altitude gain is a strong function of the single scattering albedo and the Angstrøm parameter. The smaller the Angstrøm parameter, the larger the particle. Altitude gains double from 1 to 2 km when the SSA changes from 0.9 to 0.6, and α from 2 to 1. Figure 3 demonstrates that large variations in altitude gain are possible.
3.4. Scattering, Other Absorption Agents and Cloud Albedo
 In a more general case the optical depth and single scattering albedo of an atmospheric layer are dependent on Rayleigh scattering, and the presence of cloud droplets, ozone, and water vapor. We assessed the influence of all these constituents. The presence of clouds would completely overwhelm the absorption process. Rayleigh scattering and ozone absorption both reduce the downwelling solar radiation so that their neglect exaggerates solar absorption (∼10%). Similarly, the effect of the aerosol on solar absorption is larger when the atmosphere contains less water vapor as compared to an atmosphere that contains more water vapor. The reason is that water vapor absorbs significant solar radiation, so that the effect of adding an extra absorbing constituent is diminished. We performed several simulations for layers with relative humidity varying between 0–50%, and an aerosol optical thickness of 1.75. If the baseline atmosphere (i.e., the atmosphere without aerosol) contained water vapor, then the altitude gain due to solar absorption by aerosols is about 10–15% less than situations where the baseline atmosphere contained no water vapor. Thus the drier the conditions, the larger the absorption effect of aerosols will be.
 So far, the albedo of the underlying surface has been neglected, so that illumination of the layer from below is absent. For an ocean surface with albedo smaller than 0.05 this is a realistic proposition. However, all biomass burning events initially occur over land where land-surface albedo is typically around 0.15–0.25. Thus, surface diffuse radiation will illuminate the layer from the bottom. If the layer is gradually ascending towards mid- or high tropospheric levels clouds may increase the diffuse illumination of the layer from below. Since lower tropospheric clouds have a shortwave albedo ranging from 0.3–0.6 their impact on the absorption process may be substantial. For a typical situation of τ = 1.75 at 40 degrees latitude, at 21 June the solar absorption was calculated and consequently the lifting of aerosol layers with an albedo of 0.5 as a function of SSA and α (Figure 3, broken lines). For an albedo of 0.5 between 20 to 25% additional lift can be expected.
3.5. Cross-Latitudinal Flow
 A case with an aerosol injection at 40 degrees latitude and SSA = 0.75, τ = 1.75, and α = 1.50 is used as an example to analyse a combination of potential lifting due to solar shortwave absorption and due to cross-latitudinal flow. Figure 4 shows a latitude/pressure cross section of the mean atmospheric state on 21 June as derived from the NCEP data. The coloured isopleths represent isentropic surfaces in the atmosphere (i.e., levels of equal potential temperature). Superimposed on the pressure - latitude cross section of the color coded potential temperature are the heights of the atmospheric layers [thick solid lines]. At latitudes of about 30–40 degrees the lines of equal altitude intersect with the lines of equal potential temperature at a steep angle. This means that isentropic displacement of a layer in a poleward direction (i.e., cross-latitudinal flow) results in the lifting of such a layer.
 Given the inputs of SSA = 0.75, τ = 1.75, and α = 1.50 the typical 24 hour altitude gain is 1.2 km (see Figure 3). Using an injection altitude of 3 km and remaining at 40 degrees the aerosol layer is allowed to rise to 4.2 km on day 1, and to 5.4 km on day 2 provided that there are no changes in the optical properties (Figure 4, thick white arrow). However, if a cross-latitudinal flow of 5 degrees per day is allowed, and if the layer is transported poleward the height gain is almost doubled to 6.8 km. This is due to the fact that isentropic flow near 40 degrees north on June 21 is directed towards the upper troposphere. If the flow is towards the Equator, then the situation is reversed: equatorward isentropic flow is directed towards the lower troposphere, so that it will partly offset the lifting of layers by solar heating alone. Both these situations (i.e., a 2 day 10 degree cross-latitudinal flow combined with solar heating) are illustrated in the plot by the two slanted white arrows, one ending with the tip at a latitude of 30 degrees, the other at 50 degrees north. Since many burning events originate at latitudes equatorward of 40 degrees, poleward transport is a powerful additional contributor to the ascent of aerosol layers. So, undiluted aerosol layers originating in tropical regions can reach the tropopause within 3 or 4 days simply by solar absorption and cross-latitudinal transport alone.
 In this paper we explored the potential of lifting of aerosol layers to the upper troposphere/lower stratosphere by means of solar heating. To this end we decoupled the diabatic heating due to solar heating from the adiabatic cooling due to altitude gain. Knowing the latitudinal variation of the potential temperature lapse rate, the daily diabatic heating rate of aerosol layers is translated into a tropospheric altitude gain. For optical properties typical of biomass burning events aerosol layers are lifted by 3–5 km in the course of 3 days. With additional solar heating by diffuse radiation from reflected radiation at the base of the aerosol layer and cross - latitudinal flow these gains can be doubled so that aerosol layer can quickly reach tropopause levels. Neither the presence of pyro - convection nor the onset of complex synoptic scale dynamical systems is required to allow this mechanism to achieve lift. If the aerosol layer decays as a result of sedimentation and/or chemical changes this method of vertical transport will cease to be effective. Sedimentation will remove mostly the larger particles (>10 μm) within a matter of days because of the magnitude their fall speeds. However for the remaining particles the residence time is much longer, and the higher the particles are lifted, the less the risk will be that they fall out due to precipitation which is not so relevant at higher altitudes. Thus, solar absorption can be a powerful mechanism to transport aerosol layers towards higher tropospheric altitudes.