Elevation-dependence of warming due to aerosol-induced snow darkening over the Himalayan-Tibetan region

The regional climate model (RegCM4.6) coupled with the community land model (CLM) and an online aerosol module is configured over the HT region to investigate the effects of aerosols on EDW (Giorgi et al 2012, Usha et al 2020). The model domain is shown in figure S1. The hydrostatic version of the model setup had a spatial resolution of 50 km and 18 vertical levels up to 50 hPa. For the present study, we have used Grell and Tiedtke convections, respectively, over land and ocean, the University of Washington boundary layer scheme, and the CCM3 radiation scheme. The CLM version 4.5 used in the present configuration has the Snow, Ice, Aerosol Radiative (SNICAR) module, which estimates the snow albedo variation due to changes in snow microphysical properties, aging, and the presence of impurities such as BC and dust. The online aerosol module of RegCM comprises of BC, organic carbon (OC), sulfate, dust and sea-salt aerosols. The ERA-Interim reanalysis dataset is used as the lateral and initial boundary conditions for the meteorological variables. Sea surface temperature data are adopted from NOAA reanalysis. Anthropogenic emissions of BC, OC and SO₂ were adopted from the MaCCity inventory and chemical boundary conditions from CAM/ECEARTH model simulations. The dust emission flux is estimated using the dust mobilization scheme of CLM4.5, which depends on the wind speed and surface properties. Depending on the dry deposition velocity, aerosols are continually deposited onto the surface. On the other hand, precipitation removes the particles from the atmosphere through in-cloud and below-cloud scavenging. The mixing of BC may increase radiative forcing, which would lead to an increase in the amount of radiation trapped in the snowpack (He et al 2018, Flanner et al 2021). He et al (2018) reported that the internal mixing of BC and snow increases the SAEs by 30%–60% over the Tibetan plateau, whereas changing the spherical snow grain to non-spherical snow grains decreases the albedo effect by 30%. Considering the compensating nature of the mixing state of BC and the grain shape of snow over the region, the present model configuration does not account for the mixing state and non-spherical snow shape in the simulation. Extensive validation of the meteorological and aerosol fields simulated using RegCM4.6 has been carried out with ground-based, multi-station observations and satellite data. These inter-comparison details have already been reported in several earlier publications (Usha et al 2020, 2021, 2022a, 2022b) and hence a brief review of those studies is given in table S1.

Two experiments were carried out to delineate the effects of aerosols on EDW. In the first experiment, aerosols were present in the atmosphere and snow, whereas aerosol effects were switched off in the second simulation (Control). The difference between these two simulations was used to understand the net effect of aerosols on EDW, which includes the net effect of SAE and DRE on solar and terrestrial radiation. The SAE values are estimated by running the SNICAR module in CLM4.5 for pure and polluted snow conditions. Similarly, the DRE values are obtained by the additional call of the radiative transfer module with and without aerosols in the atmosphere. The aerosol simulation used in this study includes the net effect of DRE and SAE on meteorological parameters and surface temperature. To reduce the internal variability of the model, both experiments were carried out from 2010 to 2015 (6 years) and the first year of the simulation is excluded from analysis as model spin-up. Similarly, we have carried out another set of simulations from 1990 to 1995, which were used to understand the changes in aerosol-induced surface temperature during the last two decades. Both the pentads (1991–95 & 2011–15) are marked in figure S2 along with long-term trends in temperature at 2.5 and 5.5 km. The temperature trend estimated using the CRU-TS 4.06 dataset at two different surface elevations depicts a systematic increase over the HT region, as shown in figures S2 and S3. Aerosol-induced effects on elevation dependence of surface temperature are investigated for the spring season when the high aerosol loading and solar insolation increase the natural snowmelt and hence the maximum effect of aerosols on surface temperature is observed over the entire region. To obtain the elevation dependence, aerosol and meteorological parameters are averaged at 250 m surface resolution along the HT region, and the number of model grids available for averaging at each level is shown in figure S1.

The effects of aerosol forcing on surface temperature depend on the change in the energy balance at the surface, which include changes in incoming solar radiation (R_SWin), incoming and outgoing longwave radiation (R_LWin and R_LWout) and nonradiative fluxes such as sensible (SH), latent (LH) and ground (G) fluxes (Luyssaert et al 2014, Vanden Broucke et al 2015)

$$\begin{align}\left( {1 - \alpha } \right)\,{R_{{\text{SWin}}}}\,\, - \,\,{R_{{\text{LWout}}}}\,\, + \,\,{R_{{\text{LWin}}}}\,\, = \,\,LE + SH + G\end{align} \tag{ 1 }$$

where α is the surface albedo. By converting longwave fluxes in terms of surface temperature and emissivity using the Stefan–Boltzmann equation and taking the partial derivative of the energy balance equation, we can estimate the contribution of each term to the surface temperature

$$\begin{align}\delta {T_{\text{s}}}\,\,& = \,\,\frac{1}{{4\,{\varepsilon _{\text{s}}}\,\sigma \,T_{\text{s}}^3}}\,\left[ \left( {1 - \alpha } \right)\,\delta {R_{{\text{SWin}}}}\, - {R_{{\text{SWin}}}}\delta \alpha \, + \delta {R_{{\text{LWin}}}}\right. \nonumber\\&\quad\left. - \,\delta LE\, - \delta SH - \delta G\, - \,\sigma T\,_{\text{s}}^4\,\delta {\varepsilon _{\text{s}}} \right].\end{align} \tag{ 2 }$$

Temperature variation due to (i) albedo change $\delta {T_{{\text{s}}\,\,{\text{albedo}}}}\, = \,\frac{{ - {R_{{\text{SWin}}}}\,\delta \alpha }}{{4{\varepsilon _{\text{s}}}\sigma \,T\,_{\text{s}}^3}}$, (ii) changes in shortwave flux $\delta {T_{{\text{s}}\,\,{\text{SW}}}}\, = \,\frac{{\left( {1 - \alpha } \right)\delta {R_{{\text{SWin}}}}}}{{4{\varepsilon _{\text{s}}}\sigma \,T\,_{\text{s}}^3}}$, (iii) change in longwave flux $\delta {T_{{\text{s}}\,\,{\text{LW}}}}\, = \,\frac{{\delta {R_{{\text{LWin}}}}}}{{4{\varepsilon _{\text{s}}}\sigma \,T\,_{\text{s}}^3}}$, (iv) change in sensible heat flux $\delta {T_{{\text{s}}\,\,{\text{SH}}}}\, = \,\frac{{ - \delta SH}}{{4{\varepsilon _{\text{s}}}\sigma \,T\,_{\text{s}}^3}}$, and (v) change in latent heat flux $\delta {T_{{\text{s}}\,\,{\text{LE}}}}\, = \,\frac{{ - \delta LE}}{{4{\varepsilon _{\text{s}}}\sigma \,T\,_{\text{s}}^3}}$.

The difference between the emission and deposition flux of BC aerosols clearly shows that the HT region is a major sink of atmospheric aerosols (figure 1). In contrast to South Asia and other low-elevation regions, the deposition flux of BC aerosols over the HT region is much higher than the local emission flux. This shows that anthropogenic aerosols are transported and get deposited over the HT region. This is true for most aerosol species including dust. A systematic decrease in near-surface BC concentrations with an increase in surface elevation was observed over the Himalayan region (figure 1(b)). In situ measurements of BC from stations having different elevations also showed a similar decrease in mass loading with elevation (Nair et al 2013). Compared to the Tibetan plateau, the BC values were higher for the Himalayan region due to its proximity to source regions and favourable circulation pattern. In contrast, atmospheric dust concentration slightly increased or remained almost constant with elevation. Relatively higher concentrations of dust at higher elevations are attributed to the long-range transport of dust from West Asia at high altitudes (Sarangi et al 2020).

Zoom In Zoom Out Reset image size

Compared to atmospheric concentrations, BC and dust deposited in top snow layers showed slightly increased concentrations at higher elevations due to their enrichment during the snowmelt process (figure 2(a)). The deposition of impurities like BC and dust reduces the snow albedo, which results in more absorption of solar radiation (SAE) in the snowpack (figure 2(b)). Though BC loading is several orders of magnitude lower than that of dust over the Himalayas and Tibetan plateau, the high absorption efficiency of BC results in almost comparable effects on snow albedo as that of dust over the HT region (Sarangi et al 2020, Usha et al 2020). The high values of SAE observed above 4.5 km elevation are not only due to the high concentration of aerosols but also due to the high snow depth, density and grain size at high elevations (Usha et al 2021). In the case of DRE due to the scattering and absorption of solar radiation by atmospheric aerosols, aerosol-induced surface forcing decreases with elevation (figure 2(d)), which is in line with the decrease in aerosol mass loading (figure 1(b)) and aerosol optical depth (AOD) with elevation (figure 2(c)) and also due to the increase in surface albedo as we move to higher elevation regions, where snow persists throughout the season (Nair et al 2013, Usha et al 2021). Since the same aerosol system has lower surface forcing over snow than that of the non-vegetated land, the elevation dependence of DRE has a strong dependence on aerosol loading and snow cover fraction. Though AOD is higher over the Himalayas than the Tibetan plateau, relatively higher surface albedo over the Himalayas reduces the cooling and similar DRE values are observed for both regions. At the top of the atmosphere, where the sign of the DRE is highly dependent on surface albedo, negative values of DRE change to positive values at 4.5 km over the Tibetan plateau and 4.0 km over the Himalayas respectively (figure S4). As AOD decreases with elevation, the forcing efficiency, which is forcing per unit optical depth, increased from −24 Wm⁻²AOD⁻¹ at 2.5 km to +33 Wm⁻²AOD⁻¹ at 5.5 km. In the case of DRE, the negative forcing decreases with elevation, while positive forcing due to SAE increases with elevation. Though DRE (negative forcing) and SAE (positive forcing) have compensating effects on surface forcing, both favor and strengthen EDW (Usha et al 2021).

Zoom In Zoom Out Reset image size

The energy imbalance at the surface due to atmospheric aerosols (DRE) and anomalous absorption of solar radiation by polluted snow (SAE) has implications on meteorological variables, radiative and non-radiative fluxes and snow properties (Qian et al 2011, Rahimi et al 2019, Usha et al 2020). Snow cover fraction and snow depth decrease significantly with aerosol-induced snow darkening (figures 3(a) and (b)), especially above 3.0 km over the Himalayan region. This could further enhance the absorption of solar radiation and increase the surface temperature. In the case of mountains, where snow albedo forcing is the primary physical mechanism responsible for the observed EDW, SAE could influence the warming trend significantly. As shown in figure 3(d), the aerosol forcing increases the columnar cloud cover with elevation up to 4.5 km, which decreases the shortwave radiation during the daytime, but increases the longwave during the night. The absence of change in cloud cover above 4.75 km due to aerosols could amplify the warming trend at higher elevations. Studies based on station measurements and satellite data suggest that there exists a stabilization or negative trend over the very high elevations above 5–6 km over the Himalayas and Tibetan regions (Palazzi et al 2019). This contrary could be due to the uncertainties associated with local cloud feedback of aerosol forcing.

Zoom In Zoom Out Reset image size

The combined effect of SAE and DRE on near-surface temperature shows an increasing trend with elevation (figure 4), where the increasing trend in ΔT is unambiguous. The aerosol-induced cooling regime is confined within 3.0 km (DRE > SAE) and then shifts to warming above 3.0 km elevation (SAE > DRE). The aerosol-induced local cloud feedback partially masks the warming effect of aerosols at lower elevations and enhances the warming at higher altitudes (figure 3(d)). While comparing this elevation dependence during recent (2011–2015) and past (1990–1995) pentads, an increasing trend in ΔT with elevation is clearly observed during the past two decades (figure S2). This aerosol-induced EDW could be due to the increasing trend in aerosol emissions. The emission inventories of BC over the South Asian region showed a systematic increase during the last several decades. The aerosol loading on snow is relatively higher during recent periods compared to decades earlier. This could explain the decadal increase in ΔT with elevation.

Zoom In Zoom Out Reset image size

Some studies have attributed EDW to the increase in downwelling longwave radiation (LW) (Rangwala and Miller 2012, Rangwala et al 2016). Apart from anthropogenic greenhouse gases, water vapor plays an important role in changing the longwave radiation reaching the surface. Under low humidity conditions, a small increase in the specific humidity can cause substantial longwave warming. This non-linear relationship between q and LW is suggested to play a key role in EDW during the winter season but is weak during the spring season (Rangwala and Miller 2012). Change in specific humidity due to aerosol forcing shows a clear increase (5%–15%) over the HT region (figure S5). Though the specific humidity decreases with elevation, aerosol-induced Δq shows an increasing trend with elevation. Aerosol forcing increases the specific humidity along the elevation and this increased water vapor content further increases the warming and downward longwave radiation.

To delineate the contribution of the change in (i) albedo, (ii) longwave and shortwave radiation, and (iii) sensible and latent heat fluxes in the aerosol-induced temperature change, we have used the temperature decomposition technique, which is based on the energy balance equation at the surface (Luyssaert et al 2014, Vanden Broucke et al 2015). Figure 5 shows the contribution of aerosol-induced changes to various components of the surface temperature. It should be noted that surface warming due to aerosol-induced snow darkening (δT_{s albedo}) is compensated by the decreased shortwave flux (δT_{s SW}) due to atmospheric aerosols and clouds, ground flux (δT_{s G}) and increased latent heat flux (δT_{s LH}). The sensible heat flux increases the temperature (δT_{s SH}) below 3 km and decreases the temperature above 4.5 km. The warming due to increased longwave (δT_{s LW}) is not significant compared to that of the change in surface albedo. The positive feedback between the snow albedo and surface temperature amplifies the aerosol-induced surface albedo change (Δα) and results in enhanced warming, which is significantly compensated by radiative and non-radiative fluxes. Since the aerosol forcing (surface warming due to deposition and surface cooling due to backscattering of solar radiation) depicts a systematic variation with elevation, the feedback processes, including changes in non-radiative heat fluxes, also followed similar variations with elevation. As discussed above, some of these feedback processes have negative effects on aerosol forcing (increased cloud cover and latent heat) while other feedback strengthens the forcing (increase in specific humidity and longwave). Hence, the net effect of aerosol forcing and its contribution to EDW are highly heterogeneous in space and time.

Zoom In Zoom Out Reset image size

As SAE is much higher than DRE, aerosol-induced surface warming is more prominent during spring. EDW weakens during the summer monsoon season (June–September) when most of the snow cover over the Himalayan Tibetan region melts away and albedo feedback becomes very small and confined to high-latitude mountain peaks (figure S6). Over the Tibetan region, aerosol-induced surface cooling (DRE) dominates most of the seasons except spring and winter. In contrast, aerosols warm the Himalayan region throughout the year and high values are observed during spring and summer. However, the linear trends from the 165 stations on the mean, minimum and maximum temperature over the Tibetan plateau for the years 1961–2012 report that EDW is more significant in the winter followed by spring and summer (Pepin et al 2015, You et al 2020). The changes in the shortwave and longwave fluxes over the mountain regions from the CO2-induced cloud feedback were suggested to be prominent during the winter season. In line with aerosol forcing, the forcing due to CO2 increases the cloud cover over the low-elevation regions and decreases over the high-elevation regions. This leads to the increase (decrease) in the shortwave absorption at the surface over the high (low) elevation regions resulting in enhanced EDW. However, the response of the cloud to the forcings due to CO2 or aerosols is an uncertain component that needs to be addressed in detail. Here we did not delineate the contribution of BC and dust to EDW since both have almost significant and comparable contributions to the total snow albedo reduction. In the presence of other forcing agents like CO2, the albedo feedback could be significantly amplified even at lower aerosol concentrations. Hence, aerosol forcing has a strong dependence on background meteorology and snow cover. During spring and summer seasons, the positive feedback of snow albedo processes triggered by the initial warming due to greenhouse gases was suggested as the dominant cause for the anomalous stratified warming over the high-elevation regions over the HT region (Rangwala and Miller 2012, Pepin et al 2015, You et al 2020). This warming appears to be more prominent over regions having surface temperatures close to the freezing temperature. The present study clearly shows that similar to greenhouse gases, the snow albedo feedback from the absorbing aerosols could also amplify the EDW over the HT region. Compared to the earlier reports on global warming-induced stratified trends over the HT region, aerosol-induced surface temperature changes are higher, with large spatial and seasonal heterogeneity. Positive feedback of snow-albedo is amplified with aerosol-induced snow darkening. The decrease in snow cover and shortened snow cover duration due to aerosol forcing clearly showed an elevation dependence during the spring season. EDW showed strong dependence on season, region and time period, which indicates that EDW is a regional manifestation of large-scale warming with various physical mechanisms responsible for the large heterogeneity in space and time.