A regional climate model (RegCM4.3.4) coupled with an aerosol–snow/ice feedback module was used to simulate the deposition of anthropogenic light-absorbing impurities in snow/ice and the potential radiative feedback of black carbon (BC) on temperature and snow cover over the Tibetan Plateau (TP) in 1990–2009. Two experiments driven by ERA-interim reanalysis were performed, i.e., with and without aerosol–snow/ice feedback. Results indicated that the total deposition BC and organic matter (OM) in snow/ice in the monsoon season (May–September) were much more than non-monsoon season (the remainder of the year). The great BC and OM deposition were simulated along the margin of the TP in the non-monsoon season, and the higher deposition values also occurred in the western TP than the other regions during the monsoon period. BC-in-snow/ice decreased surface albedo and caused positive surface radiative forcing (SRF) (3.0–4.5 W m-2 ) over the western TP in the monsoon season. The maximum SRF (5–6 W m-2 ) simulated in the Himalayas and southeastern TP in the non-monsoon season. The surface temperature increased by 0.1–1.5 °C and snow water equivalent decreased by 5–25 mm over the TP, which showed similar spatial distributions with the variations of SRF in each season. This study provided a useful tool to investigate the mechanisms involved in the effect of aerosols on climate change and the water cycle in the cryospheric environment of the TP.
Black carbon ; Tibetan Plateau ; Aerosol–snow/ice radiative effects ; Regional climate model
Light-absorbing impurities in snow/ice are derived from the wet and dry deposition of light-absorbing particles in the atmosphere. A few light-absorbing impurities in snow/ice can reduce ground albedo and increase the absorption efficiency of solar radiation, resulting in the melting of snow/ice at the surface. The main components of light-absorbing impurities in snow/ice are mineral dusts, black carbon (BC), brown carbon, and organic matter (OM). Mineral dusts and BC have strong absorption in the visible band, whereas brown carbon and OM absorb in the ultraviolet band. In general, mineral dusts originate from natural sources, whereas BC and brown carbon are mainly emitted from the incomplete combustion of fossil fuels and biomass during anthropogenic activities.
The Tibetan Plateau (TP), which has an abundance of snow and ice cover, is referred to as the water tower of Asia. Melting snow/ice makes a large contribution to regional hydrological resources and has direct impacts on local society and economic development. Recent studies have found that light-absorbing impurities, which may accelerate snow/ice melting, are considered as a key factor in cryospheric changes (Flanner et al., 2009 ; Doherty et al., 2010 ; Xu et al., 2009 ; Wang et al., 2013 ; Dumont et al., 2014 ). However, there have been few assessments of the radiative effects of light-absorbing impurities on snow/ice cover over the TP (Ming et al., 2009a ; Qu et al., 2014 ). Flanner et al. (2007) coupled a snow radiative model with a global climate model (GCM) and estimated the anthropogenic radiative forcing by the deposition of BC in snow averaged 1.5 W m−2 over the TP. Qian et al. (2011) also found the effects of BC and mineral dusts on changes in surface radiative forcing with the range of 5–25 W m−2 and increased the temperature by 1.0 °C on average and reduced the snowpack in spring over the TP through a GCM. However, the coarse resolution of GCMs cannot capture the spatial distribution of snow cover against observations. In this study, we used a regional climate model, which performed well in climatology over the TP (Ji and Kang, 2013 ), to simulate the concentrations and deposition of anthropogenic light-absorbing impurities (BC and OM) in snow/ice and to investigate the potential radiative effects of BC on snow/ice melting.
The Regional Climate Model version 4.3.4 (RegCM4.3.4) updated from RegCM4 (Giorgi et al., 2012 ) was used in this study. RegCM series models follow a hydrostatic equilibrium transplanted from the dynamic core in mesoscale model MM5 (Grell et al., 1994 ). The radiative transfer module is taken from the U.S. National Center for Atmospheric Research (NCAR) Community Climate Model 3.0 (CCM3) (Kiehl, 1996 ). In this study, we used the Grell (1993) convective precipitation scheme due to its good performance over these regions (Ji et al., 2015 ). The land surface module was coupled with the Community Land Model version 4.5 (CLM4.5) (Oleson et al., 2010 ). The snow module in CLM4.5 was modified by coupling it with the Snow and Ice Aerosol Radiation package (SNICAR), which can reproduce the effect of light-absorbing impurities (e.g., BC and mineral dust) on snow albedo (Flanner et al., 2007 ). In this study, the top of snow layer with the maximum thickness of 0.02 m was applied.
The initial conditions and lateral boundary conditions (ICBC) were derived from European Centre for Medium-Range Weather Forecasts (ECMWF) re-analysis ERA-interim data at 1.5° × 1.5° horizontal resolution (Dee et al., 2011 ). Sea surface temperatures were obtained from the National Oceanic and Atmospheric Administration (NOAA; Reynolds et al., 2002 ). The land cover data were obtained from the moderate-resolution imaging spectroradiometer (MODIS) (Lawrence and Chase, 2007 ). A BC and organic carbon (OC) emissions inventory (Junker and Liousse, 2008 ; Liousse et al., 1996 ) was interpolated to the model grid via a bilinear method. Emissions of OC were multiplied by 1.4 to represent OM. This OM/OC ratio is appropriate for representing fossil fuel-derived OC emissions (Russell, 2003 ). The effect of OM on snow is not considered in the current model version; therefore, only the BC in snow was investigated in this study. The model resolution was 50 km, and the domain was centered at 27°N, 85°E, with 90 and 95 grids in the north–south and west–east directions, respectively. Two simulations were performed, one with and one without aerosol–snow radiative feedback, for the period of 1989–2009 (the first year was the model spin-up). According to Wu and Zhang (1998) , we defined the monsoon season as May to September, and the non-monsoon season to be the remainder of the year.
A previous study (Ji et al., 2015 ) validated RegCM4.3 performance and confirmed that the model could reproduce the spatial distributions of atmospheric circulation, temperature, and precipitation over the TP. The model also captured the aerosol concentration and optical depth well compared with observations. Therefore, we did not repeat the model evaluation of the climatology in this study.
Fig. 1 shows BC and OM deposition on top of the snow layer based on simulations of the monsoon and non-monsoon seasons. The deposition of OM was much greater than that of BC, consistent with the difference in their atmospheric concentrations. OC emissions were 2–3 times greater than BC emissions in Asia (Ohara et al., 2007 ; Ji et al., 2015 ). In the monsoon season, BC (Fig. 1 a) and OM (Fig. 1 c) depositions were in the range of 20–120 and 60–200 μg m−2 , respectively, over the western TP and Himalayas. Aerosol deposition was very low in the inland regions of the TP. In the non-monsoon season, there were high levels of BC deposition along the margin of the Third Pole, and low levels in the inland regions (Fig. 1 b). These topographic patterns were probably associated with a terrain blocking effect on the particles in the atmosphere. In the Himalayas, Hindu-Kush, Tianshan, Kunlun, and Qilian Mountains, BC and OM depositions (Fig. 1 d) were in the range of 40–70 and 120–200 μg m-2 , respectively. In terms of seasonal differences, both BC and OM depositions were greater during the monsoon season than the non-monsoon season over the western TP. However, in the Tianshan, Kunlun, Qilian, and Himalayas, the maximum values occurred in the non-monsoon season.
Seasonal mean black carbon (BC) and organic matter (OM) deposition in snow (unit: μg m−2 ) and season mean changes of surface albedo in monsoon season (a, c, e) and non-monsoon season (b, d, f) during 1990–2009.
We summarized the mass concentrations of BC and OC in snow from 11 sites (Table 1 ) for comparison with model-simulated carbonaceous impurities on top of the snow layer. Muztagh Ata in the eastern Pamir Mountains experiences prevailing westerlies throughout the year. Meikuang Glacier is located in the eastern Kunlun Mountains, on the southern margin of the Qaidam Basin, Laohugou No. 12 Glacier, and Qiyi Glacier are located in the western Qilian Mountains, northeast of the TP. Dongkemadi Glacier is in the Tanggula Mountains in the central TP, and Lanong Glacier and Zhadang Glacier are situated in the eastern Nyainqentanglha Mountains. Qiangyong Glacier is located in the southern TP, Namunani Glacier is in the western Himalayas, and Kangwuer Glacier on Mount Shishapangma and East Rongbuker Glacier on Mount Qomolangma are located in the central Himalayas, where the monsoon climate dominates in summer. At Muztagh Ata, Qiyi, Qiangyong, Namunani, and Kangwure Glaciers, measured BC and OC concentrations were obtained from snow pits and surface snow using a two-step heating–gas chromatography method (Xu et al., 2006 ). At Meikuang, Laohugou, Qiyi (Ming et al., 2009a ), Lanong, Zhadang, and East Rongbuk, the BC mass concentrations were measured in samples from snow pits using a coulometric titration-based instrument (Ming et al., 2009a ). Another set of surface snow samples from Zhadang Glacier were measured using a thermal–optical method (Qu et al., 2014 ).
|Region||Site||Location||Elevation a. s. l. (m)||Time||BC-OBS (μg kg−1 )||BC-RCM (μg kg−1 )||OC-OBS (μg kg−1 )||OC-RCM (μg kg−1 )||OC/BC (obs)||OC/BC (RCM)||References|
|Pamir||Muztagh Ata||38.28°N, 75.02°E||6350||2001||52.1||17.6||113.2||67.9||5.4||3.9||Xu et al., 2006|
|Kunlun||Meikuang||35.67°N, 94.18°E||5200||11/2005||81.0||61.9||n.a.||171.8||n.a.||2.8||Ming et al., 2009b|
|Qilian||Lahugou||39.43°N, 96.56°E||5045||10/2005||35.0||56.3||n.a.||142.5||n.a.||2.5||Ming et al., 2009a|
|Qilian||Qiyi||39.23°N, 97.06°E||4850||6−8/2001||52.6||10.9||195.5||27.2||3.7||2.5||Xu et al., 2006|
|7/2005||22.0||n.a.||n.a.||Ming et al., 2009a|
|Ming et al., 2009a|
|Nyainqentanglha||Lanong||30.42°N, 90.57°E||5850||6/2005||67.0||17.4||n.a.||59.6||n.a.||3.4||Ming et al., 2009a|
|Nyainqentanglha||Zhadang||30.47°N, 90.63°E||5800||7/2006||114.0||17.4||n.a.||59.6||n.a.||3.4||Ming et al., 2009a|
|7−8/2012||140.0||n.a.||n.a.||Qu et al., 2014|
|South TP||Qiangyong||28.83°N, 90.25°E||5400||2001||43.1||31.8||117.3||109.9||2.7||3.5||Xu et al., 2006|
|Himalayas||Namunani||30.45°N, 81.27°E||5900||2004||4.3||19.1||51.2||69.1||11.9||3.6||Xu et al., 2006|
|Himalayas||Kangwure||28.47°N, 85.82°E||6000||2001||21.8||14.3||161.1||51.9||7.4||3.6||Xu et al., 2006|
|Himalayas||East Rongbuk||28.02°N, 86.96°E||6500||10/2004||18.0||13.8||n.a.||50.1||n.a.||3.6||Ming et al., 2009a|
The results showed that simulated BC and OC concentrations in snow were of the same magnitude as in measurements at most sites over the TP (Table 1 ) though there were some differences between observations and simulations. For example, the underestimate of BC concentration in Muztagh Ata and the Zhadang glacier, and a little of overestimated BC in the Namunani located in the western Himalayas. The available OC concentrations at the five sites were greater than those of BC, which was also reproduced in the simulations. The OC/BC ratio can be used to approximately indicate the source of particles. As for the TP, the mean OC/BC in emissions derived from biomass and fossil fuel burning were 6.9 and 2.7 (Streets et al., 2003 ), respectively. When the OC/BC ratio is much higher (lower), it suggests a greater contribution of carbonaceous impurities in snow/ice from biomass (fossil) fuel burning. The largest OC/BC ratios were measured at Namunani and Kangwure, which were considered to be remote areas with few local anthropogenic emissions. In contrast, the low values of the OC/BC ratio at Laohugou and Qiangyong indicated that carbonaceous material from local fossil fuel combustion was probably deposited on the glacier. Compared with measurements, the model produced similar OC/BC ratios in the northeastern (Qiyi) and southern (Qiangyong) TP; however, it underestimated the ratio in the Himalayas (Namunani and Kangwure) and Pamir Mountains (Muztagh Ata). This suggests that the model could reproduce the contributions of anthropogenic impurities in snow/ice better than it could reproduce the regional background. However, it should be noted that there were large uncertainties in the OC/BC ratio due to local effects and chemical processes, and the implications of the OC/BC ratio could not be reliably derived due to the lack of long-term continuous measurements.
The darkening of snow cover by the deposition of BC changed the surface albedo, reducing it by 0.04–0.10 over the western TP and reducing that in the Tianshan and Kunlun Mountains by 0.01–0.04 during the monsoon season (Fig. 1 e). In the non-monsoon season, the surface albedo decreased by 0.04–0.08 along the mountainous regions surrounding the Third Pole (Fig. 1 f), while there was a lesser reduction in surface albedo (range of 0.01–0.04) in the inland regions. The maximum decreases in surface albedo (0.08–0.12) occurred over the western Himalayas during the monsoon season. In the northern and eastern TP, the largest reductions occurred during the non-monsoon season.
In this study, the radiative forcing was defined as an instantaneous change in net radiative fluxes (down minus up) induced by BC-in-snow. The surface radiative forcing (SRF) caused by darkened snow was positive, with values in the range of 0.0–4.5 W m−2 over the TP during the monsoon season (Fig. 2 a). The maximum SRF, with values in the range of 3.0–4.5 W m−2 occurred in the western Himalayas and eastern Pamir Mountains. In the Tienshan, Kunlun, and Qilian Mountains and in the eastern TP, the SRF was in the range of 0–3 W m−2 . During the non-monsoon period (Fig. 2 b), the SRF was larger, with positive values of 5–6 W m−2 in the Himalayas and southeastern TP, whereas in the inland regions of the TP and Tianshan Mountains, the SRF was 1.0–3.5 W m−2 , and it exhibited a topographic pattern consistent with the concentration of carbonaceous particles in snow. Our SRF results were within the range reported in a previous study based on a GCM, which estimated the SRF caused by BC in snow to be in the range of 0.5–8.5 W m−2 (Flanner et al., 2009 ) over the TP in spring. Ming et al. (2013) calculated the area-averaged radiative forcing via BC in snow as 2.9–10.3 W m−2 at 17 sites over the TP.
Seasonal mean changes in surface radiative forcing (a, unit: W m−2 ), temperature (c, unit: °C) and snow water equivalent (SWE) (e, unit: mm) in the monsoon season during 1990–2009 (b, d, f as a, c, e but for the non-monsoon season).
As a response to SRF, the 2-m temperature increased by 0.1–1.5 °C over the Tianshan, Pamir, and the Himalayas during the monsoon season. Maximum warming, in the range of 1–1.5 °C, occurred over the eastern Pamir Mountains and western Himalayas. However, the changes in temperature in the central and eastern TP were not significant. During the non-monsoon season, temperature also increased due to the darkening of snow over the Third Pole regions. Considerable warming (exceeding 1.5 °C) was evident in the western TP, with a smaller temperature increase (1.0–1.5 °C) occurring in the southeastern TP. Because of the warming effects induced by particles in snow, the snow water equivalent (SWE) decreased by 5–25 mm over the western TP and Himalayas during the monsoon season. The maximum reductions in SWE were 10–25 mm in the western Himalayas. In the Tianshan and Kunlun Mountains, the SWE decreased by 1–5 mm. The decrease in the SWE during the non-monsoon season was greater than that during the monsoon season and covered most regions of the Third Pole. The greatest reductions in the SWE (10–25 mm) also occurred over the Himalayas and western TP.
Light-absorbing impurities in snow/ice have a large impact on cryospheric changes; however, few studies have focused on their radiative effects over the TP. This study simulated the deposition of anthropogenic light-absorbing impurities in snow/ice and assessed the radiative impact of BC over the TP using RegCM4.3.4–CLM4.5. The results indicated that the model performed reasonably well in terms of the spatial distribution of BC and OM deposition and their surface concentrations. The deposition fluxes of BC and OM were greater in the west than in the other regions of the TP during the monsoon season. High values of BC and OM deposition were also found along the margin of the TP due to topographic blocking in the non-monsoon season. The total deposition of BC and OM at the surface during the non-monsoon season was greater than that during the monsoon season over the TP.
The presence of BC caused a reduction in surface albedo and positive SRF over the TP during both the monsoon and non-monsoon seasons. The radiative forcing from BC in snow was 3.0–4.5 W m−2 in the western TP and Himalayas in the monsoon season. During the non-monsoon season, the maximum radiative forcing (5–6 W m−2 ) occurred in the Himalayas and southeastern TP. These radiative effects had a similar spatial distribution to the BC concentrations in snow during each season. Due to the positive radiative forcing induced by BC in snow, surface temperature increased by 0.1–1.5 °C and SWE decreased by 5–25 mm over the TP, with the largest variations in the Himalayas and western TP. In a previous study, RegCM4.3.4 was validated as a useful tool to simulate mineral dusts over High Mountain Asia (Ji et al., 2016 ). However, there were several challenges in modeling anthropogenic light-absorbing impurities in snow/ice. First, the OM emission inventory was obtained from a constant OC emission flux; however, the approximate OM/OC ratio did not represent biomass burning of OM from OC emissions. As a result, there may be large uncertainties in the OM deposition in this study. In the current model revision, the radiative effect of OM was not included, as some previous studies suggested that the optical effects of OM in snow are very small (Hess et al., 1998 ; Qian et al., 2011 ). Second, the mineral dusts which were considered as the great components of light absorbing impurities were not involved in current study. Also, the impact of light absorbing impurities on glaciers was not considered in the model. Third, the parameterization of the aerosol–snow feedback in the coupled model could not represent the real situation in the TP in terms of particle sizes and shapes and snow grains due to the limited number of samples and laboratory analyses undertaken. The Atmospheric Pollution and Cryospheric Change (APCC) project will focus on the impact of atmospheric pollution on climatic and environmental changes in the cryosphere, and will build a network to address atmospheric aerosols, snow pit sampling, and glacier monitoring over the TP, Antarctic, and Arctic. The datasets from this program will optimize model parameterization and improve our understanding of the effects of light-absorbing impurities in snow/ice on cryospheric change.
This study is supported by National Nature Science Foundation of China (41301061 ), Chinese Academy of Sciences (KJZD-EW-G03-04 ) and China Meteorological Administration Special Public Welfare Research Fund (GYHY201306019 ).