Climate Dynamics

, Volume 36, Issue 9, pp 1633–1647

Simulation of the anthropogenic aerosols over South Asia and their effects on Indian summer monsoon

Authors

  • Zhenming Ji
    • Key Laboratory of Tibetan Environment Changes and Land Surface ProcessesInstitute of Tibetan Plateau Research, Chinese Academy of Sciences
    • National Climate Center
    • Graduate University of Chinese Academy of Sciences
    • Key Laboratory of Tibetan Environment Changes and Land Surface ProcessesInstitute of Tibetan Plateau Research, Chinese Academy of Sciences
    • State Key Laboratory of Cryospheric ScienceChinese Academy of Sciences
  • Dongfeng Zhang
    • Shanxi Meteorological Bureau
  • Chunzi Zhu
    • College of Atmospheric ScienceNanjing University of Information Science Technology
  • Jia Wu
    • National Climate Center
  • Ying Xu
    • National Climate Center
Article

DOI: 10.1007/s00382-010-0982-0

Cite this article as:
Ji, Z., Kang, S., Zhang, D. et al. Clim Dyn (2011) 36: 1633. doi:10.1007/s00382-010-0982-0
  • 358 Views

Abstract

A regional climate model coupled with a chemistry-aerosol model is employed to simulate the anthropogenic aerosols including sulfate, black carbon and organic carbon and their direct effect on climate over South Asia. The model is driven by the NCAR/NCEP re-analysis data. Multi-year simulations with half, normal and double emission fluxes are conducted. Results show that the model performs well in reproducing present climate over the region. Simulations of the aerosol optical depth and surface concentration of aerosols are also reasonable although to a less extent. The negative radiative forcing is found at the top of atmosphere and largely depended on emission concentration. Surface air temperature decreases by 0.1–0.5°C both in pre-monsoon and monsoon seasons. The range and intensity of cooling areas enlarge while aerosol emission increases. Changes in precipitation are between −25 and 25%. Different diversifications of rainfall are showed with three emission scenarios. The changes of precipitation are consistent with varieties of monsoon onset dates in pre-monsoon season. In the regions of increasing precipitation, monsoon onset is advanced and vice versa. In northeast India and Myanmar, aerosols lead the India summer monsoon onset advancing 1–2 pentads, and delaying by 1–2 pentads in central and southeast India. These changes are mainly caused by the anomaly of local Hadley circulations and enhancive precipitation. Tibetan Plateau played a crucial role in the circulation changes.

Keywords

Indian summer monsoonRegional climate modelAerosolsClimate effect

1 Introduction

Aerosols play an important role in both the global and regional climate balance. They are recognized as a significant climate forcing, a factor that alters the earth’s radiation balance and thus possibly affecting climate. In recent decades, India as the biggest country in South Asia, emissions of anthropogenic aerosols rank the second largest nation in Asia (Ohara et al. 2007), which inevitably has an impact on the climate of South Asia.

South Asia climate is mainly influenced by the Indian monsoon and Westerlies with clear four seasons: winter (January and February), a pre-monsoon season (March to May), a summer monsoon (rainy) season (June to September), and a post-monsoon period (October to December).

The aerosol’s direct effect is the mechanism by which aerosols scatter and absorb shortwave and longwave radiation, thereby altering the radiative balance of the earth-atmosphere system. Sulfate, fossil fuel organic carbon (OC) and black carbon (BC) are significant anthropogenic aerosols. Sulfate is an entirely scattering aerosol across the solar spectrum but with a small degree of absorption in the near-infrared spectrum. Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (IPCC 2007) estimates the sulfate aerosol direct radiative forcing is −0.4 ± 0.2 W/m2. OC aerosols are emitted as primary aerosols or formed as secondary aerosols from condensation of organic gases considered semi-volatile or low volatile. OC aerosols from fossil fuels are absorbing at some ultraviolet and visible wavelengths. Their direct radiative forcing is −0.05 ± 0.05 W/m2 (IPCC 2007). BC is produced when biomass or fossil fuel is insufficiently combusted. It absorbs solar radiation strongly in the visible and infrared. Thus, although its concentration in the atmosphere is low, it greatly affects solar radiation absorption and rises the temperature of the atmosphere, significantly, altering the regional and global climate.

Researches have been conducted on the effect of anthropogenic aerosols to climate by using the climate models over this region (Collins et al. 2002; Menon et al. 2002; Meehl et al. 2008; Ramanathan et al. 2005; Ramanathan and Carmichael 2009). These studies have provided some significant results including radiative forcing caused by aerosols and their impact on temperature and precipitation. However, IPCC AR 4 has pointed out that aerosol effects on climate had a very large uncertainty. Although the model results in this region were consistent in the qualitative aspects, there was a big difference of quantities among each other. In addition, those works didn’t consider common role of multiplicate aerosols. Sulfate and OC are both scattering aerosols, while BC is absorbing aerosol, and their combination is not a simple linear superposition, but depends on three kinds of aerosols mixed radiation.

Lau et al. (2006) used NASA finite-volume GCM to investigate a plausible mechanism for aerosol impact on the Asian summer monsoon. They suggest that increased dust loading coupled with black carbon emission from local sources in northern India during late spring may lead to an advance of the rainy periods and subsequently an intensification of the Indian summer monsoon. Liu et al. (2009) utilized CAM 3.0 examining the direct effects of sulfate and BC aerosols in China on East Asia monsoon. They found that the mixed aerosol weaken East Asia monsoon in both summer and winter seasons. However, all above-mentioned studies have used the global climate model, which has a lower resolution and was difficult to capture subtle characteristics in the areas of complex terrain.

The advantages of high-resolution regional climate model can compensate for the defect of a global model in this regard. Studies have shown that high-resolution regional climate model had better simulation capabilities than the global model for the monsoon climate in Asia (Gao et al. 2006, 2008; Zhou and Yu 2006).

Several researches (Solmon et al. 2006; Zakey et al. 2006) based on the RegCM-chemistry model to simulate the aerosols over monsoon region of Africa. The conclusion indicated that the coupled model could well reproduce the regional pattern of climate and aerosols’ distribution. However, the previous studies largely focus on the testing model performance, and short of the aerosol effect on monsoon climate.

We used this RegCM-chemistry model to simulate the temporal and spatial distribution of dominating anthropogenic aerosols (sulfate, BC, OC) over Asia, prior to analyzed the impact of aerosols on climate in China (Ji et al. 2010). In this work, we focus on anthropogenic aerosols in South Asia and their effects on Indian summer monsoon.

2 Model and experiment design

The model employed is the Regional Climate Model version 3.0 (RegCM3) developed at the Abdus Salam International Center for Theoretical Physics (Giorgi and Mearns 1991; Giorgi et al. 1993a, b; Pal et al. 2007). Radiative transfer module is from the NCAR global model CCM3 (Kalnay et al. 1996). Land surface processes are represented via the Biosphere–Atmosphere Transfer Scheme (BATS1e; Dickinson et al. 1993). The planetary boundary layer computation is employed the non-local formulation of Holtslag et al. (1990). In these experiments, we chose Grell convective precipitation scheme with Fritsch-Chappell type closure (Grell 1993).

The coupled aerosol model is developed by Solmon et al. (2006) based on the scheme of Qian et al. (2001), which contains advection, horizontal and vertical turbulent diffusion, cumulus convective transportation, surface emission, wet removal by large-scale and convective rain, dry deposition, production and losses due to physico-chemical transformations. In considering the difference between radiative properties of various aerosols, the model parameterizes mass extinction coefficient, single scattering albedo and asymmetric factor of per unit mass of the various aerosols (sulfate, hydrophilic BC and OC, hydrophobic BC and OC) at different wavelengths (19). When calculating the optical properties of sulfate, hydrophilic BC and OC aerosols, the model also consider the effects of humidity. Different types of aerosols use different fitting formula to calculate their hydrophilic growth (Solmon et al. 2006).

Aerosols affect climate in two ways: directly, through scattering and absorption of solar and infrared radiation, and indirectly, by influencing the cloud optical properties and lifetime (Liou 2002). In this study, only the direct effect is included.

Four experiments were designed in this work. The control experiment (CON) does not include aerosols; the other three sensitivity experiments contain three anthropogenic aerosols (sulfate, BC, OC) with normal (SEN1), half (SEN2), and double (SEN3) emission quantities. CON and SEN1 are used to evaluate the model performance. The differences between SEN1 and CON are considered the present climate effects of aerosols. The experiments with half (SEN2) and double (SEN3) aerosols are tested for sensitivity of aerosols effect on summer monsoon climate.

Horizontal resolution is 50 km for each simulation. Central point of the model is set at 32°N, 99°E, with 160 grids in the north–south direction and 192 grids in the west-east direction. The model is run at its standard configuration of 18 vertical sigma layers with the model top at 10 hPa.

The simulation period is from January 1, 1987 to December 31, 2009, for a total of 23 years. The first 12 months is considered as a model initialization (spin-up) time and is not analyzed. In general definition, we analyze pre-monsoon season which is from March to May (MAM), and monsoon season which is June-July–August–September (JJAS). Figure 1 shows the model domain and topography. It includes the major emission countries, such as China and India. The box is the key region that we special analyzed.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig1_HTML.gif
Fig. 1

Model domain, topography (units: meter)

Initial and lateral boundary conditions are obtained from the NCAR/NECP re-analysis data (Kalnay et al. 1996). Sea surface temperature used comes from National Oceanic and Atmospheric Administration (NOAA; Reynolds et al. 2002). Land use types are based on observed data within China (Liu et al. 2003) and satellite GLCC (Loveland et al. 2000) data developed by the USGS outside China.

The emission inventory is derived from REAS (Regional Emission inventory in Asia; Ohara et al. 2007) produced by Frontier Research Center for Global Change (FRCGC). The data set only has inter-annual change and is at a resolution of 0.5° × 0.5°. We interpolate the data to the model grid point by using a bilinear interpolation method. Figure 2 shows the average emission flux of REAS (SO2, BC, OC) from 1988 to 2009, and similar characteristics can be found, with high value in India and low value in Tibetan Plateau. In general, the value of SO2 is larger than BC and OC by an order of magnitude.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig2_HTML.gif
Fig. 2

Emission inventory in South Asia (units: 10−10 kg m−2 s−1): a SO2; b BC; c OC

The datasets from Climate Research Unit (CRU; Mitchell and Jones 2005), the NCAR/NECP re-analysis and Xie-Arkin (Xie et al. 2007) are used to validate the simulated the surface air temperature, atmospheric circulation and precipitation, respectively. We use Moderate-resolution Imaging Spectro-radiometer satellite (MODIS; http://modis.gsfc.nasa.gov/), Multi-angle Imaging Spectro-radiometer satellite (MISR; http://www-misr.jpl.nasa.gov/) and Aerosol Robotic Network (AERONET; http://aeronet.gsfc.nasa.gov/) datasets to compare with simulated aerosol optical depth (AOD). The near surface concentrations of black carbon from other literatures are collected for testing the model output.

3 Results

3.1 Model performance

The model ability of representing regional climate is the basis to carry out simulation aerosols and estimate their climate effect by using RegCM3. Previous studies have indicated that RegCM3 had a good capability to capture the characteristics of the Indian summer monsoon climate (Dash et al. 2006; Ashfaq et al. 2009).

The observed and simulated surface air temperature for both pre-monsoon and monsoon season are shown in Fig. 3. The simulated surface air temperature shows generally good agreement with observations. India’s climate is strongly influenced by the Himalayas in the north and the Thar Desert in the northwest. The Himalayas act as a barrier to the frigid katabatic winds flowing down from Central Asia. Thus, in pre-monsoon, the subcontinent is relatively and homogeneous hot with mean temperature beyond 24°C (Fig. 3a, b). In monsoon season, the Thar Desert with a high temperature center in northwest India and southeast Pakistan (Fig. 3c, d). However, there is a cold bias of 1–3°C in central, northeast and south India subcontinent.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig3_HTML.gif
Fig. 3

a, c Observed and, b, d simulated mean temperature (units: °C) in pre-monsoon (a, b) and monsoon season (c, d)

Figure 4 depicts observed (NCEP re-analysis) and simulated mean 850 hPa wind and precipitation both in pre-monsoon and monsoon season, respectively. In pre-monsoon season, northwest current is divided into two branches due to the Deccan Plateau (Fig. 4a). One branch flows east, and converges with south wind from the Bay of Bengal. Finally, the southwesterly flow covers northeast India, northern Bay of Bengal, and Myanmar. The other turns to south, and towards the Arabian Sea. In monsoon season, there are prevailing southwest winds from the Bay of Bengal to low latitudes areas near the equator (Fig. 4c). West winds dominate in other region except the lower wind velocity in the higher latitude. Comparing with the observations, the model simulates well in pre-monsoon season (Fig. 4b). But in monsoon season, the deviation of simulation is a cyclone in North India (Fig. 4d). It is a possibility that the re-analysis may have bias itself. The global climate model is introduced by the lower resolution (T63) in the steep mountains (Himalaya). Gao et al. (2008) find that the finer resolution model inhibits the penetration of the monsoon precipitation front from the southern slope of the Himalaya. Here, we also consider the presentation of cyclone in North India is possibly due to the high resolution RegCM introduces a fine scale topographically-induced structure.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig4_HTML.gif
Fig. 4

a, c Observed and b, d simulated mean precipitation (units: mm) and 850 hPa wind (units: m/s) in pre-monsoon (a, b) and monsoon season (c, d)

Meanwhile, the model can simulate the basic position of the rain band in South Asia. In pre-monsoon season, the precipitation is low in Indian subcontinent (Fig. 4a). Large areas are less than 250 mm. Precipitation impacted by terrain is also clear in monsoon season (Fig. 4c), such as in India’s southwest coast, the precipitation is blocked by the Deccan Plateau, and mostly concentrates in the mountains of the windward side. In Tibetan Plateau, the model could reproduce the diminishing rain belt from southeast to northwest. Compared with the observations, the simulated precipitation is underestimated in northwest and central India (Fig. 4b) and overestimated in northeast India in pre-monsoon season. In JJAS, the model simulated a lower rainfall value in the Northwest semi-arid areas (Fig. 4d). The greatest underestimate of precipitation is found in Northeastern India and Myanmar along the Bay of Bengal.

It is generally considered that the Asian summer monsoon firstly break out in early May in South China Sea, and then gradually advances to the northwest. In early June, it arrives to the Bay of Bengal. Finally, the India monsoon is outbreak (Tao and Chen 1987). A lot of indexes are expressed the summer monsoon onset. Hydrologic onset and withdrawal index from Fasullo and Webster (2003) are considered practical. However, it is focused on interannual variability of monsoon transitions. We use another simple index of climatological monsoon onset which is defined as the pentad (5-day mean) precipitation is exceeded 5 mm/day and the January mean value (Wang and LinHo 2002; Ashfaq et al. 2009).

We give the timing of Indian summer monsoon onset from observations (Xie et al. 2007; Fig. 5a) and model simulation (Fig. 5b). Indian summer monsoon is breaking out from the 31st pentad (early June) in India, and gradually move to the northwest. In the 39th pentad (later July), it arrives to Afghanistan and Pakistan. On the other side of the Bay of Bengal, the countries such as Burma and Thailand, the monsoon onset is earlier than the Indian subcontinent. From the comparison of the observation and simulation, the model could well represent the timing of India summer monsoon onset. Ashfaq et al. (2009) use CMAP dataset (observation) and RegCM (nest FvGCM) output to analyze the timing of the Indian summer monsoon onset. Our conclusion is consistent with their results.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig5_HTML.gif
Fig. 5

a Observed and b simulated India summer monsoon onset (units: pentad)

Figure 6a–c shows the AOD from observations and SEN1 during the monsoon season. The comparison suggests that the model captures the areas of low values center in the southern Indian Peninsula and Tibetan Plateau, as well as high values in central and eastern India. However, the simulated values lower than satellite data over entire range of South Asia, particularly in Northern India and Pakistan. It is mainly because the AOD estimated from MODIS and MISR are including total aerosols, whereas our study only includes three types of anthropogenic aerosols (no mineral dust aerosols, thus the simulation is generally lower than observations, the deficiency should be largely reflects the composition of dust aerosols). The Thar Desert in Northwest generates a large number of dust aerosols and dominates this location. Pandithurai et al. (2008) indicate that an increased abundance of coarse particles in New Delhi due to dust storms that transport desert dust from the Thar Desert, and the a consistent increase in aerosol loading from March to June with monthly average AOD values at 0.5 mm of 0.55, 0.75, 1.22 and 1.18, respectively, was observed. Although their study only focuses on the year 2006, it can be seen that the contribution dust aerosol to the AOD is very high in northwest India. Meanwhile, the Gangetic Plains, bounded by the high altitude Himalaya Mountains, in the central and northern India, are strongly influenced by the transport of dust outbreaks of the Thar Desert in the northwestern India (Gautam et al. 2009).
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig6_HTML.gif
Fig. 6

Mean aerosol optical depth from a, b observed and c simulated

We select six sites (mark from 1 to 6, Fig. 7) from AERONET in this region and compare AOD values with SEN1 output (by 90% confidence level). These are the only available sites in South Asia with continuous AOD values. As Table 1 shows, at Kanpur which is mainly affected by dust aerosol (Singh et al. 2004), the monthly (12 months) correlation coefficient between observation and simulation is merely −0.27. In monsoon season, the mean observed AOD is 0.55, while model simulated is 0.25. Site of Gandhi College, a relatively less influence by dust aerosol, the correlation coefficient reached 0.48 and the error is 18%. In the northern highland, EVK2-CNR and Nam Co sites which are less impressionable than dust-affected zone, the correlation coefficient between observed AOD and simulated value is much greater. In the eastern Bay of Bengal, Chulalongkorn site in Thailand, the correlation coefficient is 0.45. The observation is 0.13 and the simulation is 0.11 in Silpakorn site in monsoon season, but the correlation coefficient is only −0.24. Nevertheless, very few observation data limit our comparisons and further validation of simulated results.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig7_HTML.gif
Fig. 7

Six stations (16) from AERONET and eight sites (AH) collected from other literatures

Table 1

Observed (AERONET) and simulated AOD for monsoon season

Location

Time

Monthly

Monsoon AOD

Correlation coefficient

Aeronet

RegCM3

1 (26.51°N, 80.23°E)

2001–2008

−0.27

0.55

0.25

2 (25.87°N, 84.13°E)

2006–2008

0.48

0.46

0.39

3 (27.96°N, 86.81°E)

2006–2007

0.89

0.11

0.11

4 (30.77°N, 90.96°E)

2006–2007

0.57

0.05

0.04

5 (13.82°N, 100°E)

2006–2009

−0.24

0.13

0.11

6 (13.73°N, 100.5°E)

2003–2004

0.45

0.28

0.12

In addition, we compare the BC surface concentration from eight observation sites (mark from A to H, Fig. 7) and model simulation (Table 2). We note that the model underestimate in seven samples. At least parts of the reasons for these differences are owing to a problem of point source representation at the 50 km model grid, but observation site is only a spot nearly several square meters. And the later is easily influenced by weather condition, terrain, constructions and human activities. In other five samples, the model is more specifically representing the observations.
Table 2

Observed and simulated BC surface concentration (units: μg/m3) based on a number of studies reported in the literature

Location

Time

BC concentration

Refs.

OBS

RegCM3

A (29.4°N, 79.5°E)

2004/12

1.36 ± 0.99

1.11

Pant et al. (2006)

B (26.4°N, 80.3°E)

2004/12

6–20

3.77

Tripathi et al. (2005)

C (8.55°N, 77°E)

2000/12–2001/03

5

1.45

Babu and Moorthy (2002)

 

2001/06–2001/09

1.5

1.42

 

D (17.5°N, 78.4°E)

2003/05

0.5–68

2.24

Latha and Badarinath (2003)

 

2003/07

0.5–45

2.59

 

E (19.4°N, 72.8°E)

1999/01–1999/02

12.4 ± 5.1

1.81

Venkataraman et al. (2002)

 

1999/02–1999/03

12.6 ± 3.0

1.74

 

F (13°N, 77°E)

2001/10–2001/12

4.2 ± 0.03

1.10

Babu et al. (2002)

G (15.45°N, 73.8°E)

1998/12–1999/02

3 ± 0.7

1.38

Leon et al. (2001)

 

1999/03

1.5 ± 0.7

1.53

 

H (15.42°N, 74.98°E)

1999/02

3

1.25

Leon et al. (2001)

RegCM3 and its chemical processes are developed on the climatic characteristics of Europe (Solmon et al. 2006). After several years of application and development in Asia, RegCM3 has updated climate simulation capability in this area (Dash et al. 2006; Gao et al. 2008; Ashfaq et al. 2009), but it significantly lacks in the aerosol simulation and also needs for update in future.

3.2 Aerosol effect on Indian summer monsoon climate

The aerosol radiative forcing is usually looked upon one of causes that change the radiation balance of the earth-atmosphere system. Figure 8a, b shows the distribution of simulated aerosol short-wave radiative forcing in the top of atmosphere (TOA) with present emission. We find the mixed aerosol radiative forcing is negative. It is mostly because the scattering aerosols (e.g. sulfate and OC) dominate this region (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig8_HTML.gif
Fig. 8

Simulated aerosol short wave radiative forcing (units: W/m2) at the top of atmosphere in pre-monsoon (a, c, e) and monsoon season (b, d, f) for SEN1 (a, b), SEN2 (c, d), SEN3 (e, f)

In per-monsoon season, the major negative district at the top of atmosphere is along the Bay of Bengal (Fig. 8a). In Bangladesh and northeastern India, the center value exceeds −3 W/m2. In monsoon season, the negative distribution is broader than the pre-monsoon (Fig. 8b), including the Tibetan Plateau, Indochina peninsula and India. Maximum negative area is located in northern India with the center value beyond −6 W/m2. Qian et al. (2003) used the MM5 simulated climatic effects of aerosols during 5 years around China. Comparing with their results, the radiative forcing at the top of atmosphere has a similar distribution pattern and values in northeastern India and Indochina peninsula. Figure 8c–f display the distribution of TOA of SEN2 and SEN3 respectively. It is clearly that the radiative forcing is significantly correlated with the concentration of emissions. If emission flux increases, the intensity and range of negative radiative forcing would enlarge. The distribution of radiative forcing at the surface is alike the the top of atmosphere, but values are much larger (figures are not showed).

Figure 9 shows the aerosol effects on temperature and 850 hPa wind field. With the normal aerosol emission flux (SEN1), the surface air temperature decreases by 0.1–0.5°C over the northern India and Myanmar (Fig. 9a) in pre-monsoon season. Cooling in the north-central India is due to the higher aerosol emissions and the synthetic aerosol lead negative shortwave of surface radiative forcing. Aerosol emission in Myanmar is not high, but there are also significant cooling areas.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig9_HTML.gif
Fig. 9

Simulated aerosol effects on temperature (units: °C) and 850 hPa wind field (units: m/s) in pre-monsoon (a, c, e) and monsoon (b, d, f) season for SEN1 (a, b), SEN2 (c, d), SEN3 (e, f)

Changes in 850 hPa wind field note that the westerlies in the low latitude area over Arabian Sea, are across the southern India and ultimately arrive to Bay of Bengal. Then they gradually turn to northeast. The variations of currents are mainly due to the change of pressure. Changes in 850 hPa geopotential height field show that there is a high negative center in the west Bay of Bengal. The center exceeds −1 geopotential meter, and exists with a warming air mass. The cyclonic circulation extends to the low-latitude regions, so that tropical area over Arabian Sea and Bay of Bengal which are in the south of cyclone are prevailing westerly wind. The southwest wind flows over the northeastern Bay of Bengal lead to an increase in regional precipitation. The increased precipitation also possibly cools the surface air temperature.

In monsoon season, the cooling areas are in the northwestern Indian subcontinent (Fig. 9b) and the surface air temperature decreases by 0.1–0.5°C. The center is colder than −0.5°C. Myanmar and northeastern India is cooling by 0.1–0.5°C.

There are strong westerlies from the Arabian Sea in the south of 20°N. We find a cooling center in the Northwest. The lower columns are condensed and air pressures are increased. So a positive geopotential height anomaly generates in the Northwest. For compensating the losses in the high layers of anti-cyclone, a cyclonic circulation appears in the south of it. Westerlies in the south of cyclone dominate the region below 20°N.

The cooling areas also present over the entire region when the aerosol emissions reduce by half (Fig. 9c, d). But the magnitude and range of cooling reduce comparing with the normal emission. In pre-monsoon season, westerly winds are weakened in tropical region. Meanwhile, cooling area and southwest winds in northeast Bay of Bengal have disappeared. Figure 9e, f show the changes about double emission flux. Surface air temperature decreases significantly over the entire South Asia subcontinent both in the pre-monsoon and monsoon seasons. The changes are larger than SEN1’s conclusion. Different from the SEN2, west winds are much stronger and control the region of the Bay of Bengal. As a result, they bring more warm and moist air to these areas.

It is complex about varieties of precipitation. In the north India and Indochina Peninsula, the rainfall increases by 5–15% in pre-monsoon (Fig. 10a). But it reduce 5–25% in south India. As above mentioned, the increase precipitation along the Bay of Bengal is partly because of the abundant water vapor flowed by southwesterlies from moist ocean.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig10_HTML.gif
Fig. 10

Simulated aerosol effects on precipitation (units: %) and 850 hPa wind field (units: m/s) in pre-monsoon (a, c, e) and monsoon (b, d, f) season for SEN1 (a, b), SEN2 (c, d), SEN3 (e, f)

Changes in precipitation (Fig. 10b) are similar to these of temperature in monsoon season. It increases by 10–25% in the Northwest and Tibet Plateau. The regions with enhanced precipitation extend eastward along the southern foot of the Himalayas in northeast India. In central and western parts of the coast, the precipitation decreases by 5–10%. The cooling of Northwestern India is as a result of the high sulfate emission (Fig. 2a). It also might be related with the increasing precipitation in this region. According to changes of the 850 hPa wind field, a cyclonic circulation presents in the north-west Indian subcontinent under the influence of aerosols, which leads the upward motion and increasing precipitation. Contrarily, the central and western regions are under the control of anti-cyclonic circulation. The sinking motion would strengthen the stability of the atmosphere, resulting in reduced precipitation.

Rainfall increased areas in Myanmar disappear when the aerosol emission reduce by half (Fig. 10c). That is probable due to the weakened southwesterlies in pre-monsoon season. In monsoon season, the enhanced precipitation concentrates in the northwest India and Pakistan (Fig. 10d), and the center value is larger than 25%. When the emissions increase by double, changes in are different from SEN1 and SEN2. Although the variational values are basically between −25 and 25%, the strengthened precipitation is more extensive in the northeast Bay of Bengal during pre-monsoon season (Fig. 10e). In addition, rainfall increases by 5–10% in the west coast of India (Fig. 10e, f).

As can be seen from the above conclusion, aerosols’ effects on temperature are more direct than precipitation. It is mainly because the model just included direct effects of aerosols which straight disturb radiative forcing and energy balance, and directly affect the temperature changes. Meanwhile, the diversification of atmospheric temperature makes the air pressure circulation anomalies, and finally changes precipitation.

3.3 Aerosol effect on Indian summer monsoon onset

Figure 11a shows the change of Indian summer monsoon onset (SEN1 minus CON). The monsoon onset dates over the central and southeast India delay by 1–2 pentads. Myanmar and some region of northeast India are ahead about 1–2 pentads. Tibetan Plateau, Pakistan and northwest India have little change. Change in precipitation (Fig. 10a) is consistent with the Indian summer monsoon onset. In the enhancive precipitation areas, the monsoon breaks out in advance. Vice versa, the monsoon onset is delayed in the reduced precipitation regions. Monsoon onset advances 1–2 pentads over the rare zone of western India with SEN2 (Fig. 11b). In SEN3 (Fig. 11c), the onset changes are different from the normal and half emission flux. Along coast in southwest India and the Bay of Bengal, the onset dates are ahead by 1–2 pentads. However, in eastern and central India, the dates delay between one and two pentads.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig11_HTML.gif
Fig. 11

RegCM3 simulated changes (a SEN1-CON; b SEN2-CON; c SEN3-CON) in summer monsoon onset dates (units: pentad)

We analyze possible cause about the changes of monsoon onset with normal emission. In pre-monsoon season, there is a larger concentration of black carbon aerosols in South Asia. The absorption of solar radiation by BC heats up the atmosphere and ulteriorly impacts on circumfluence. Fig. 12a displays the differences (SEN1-CON) of latitude-height cross-section for mean temperature (shade) and meridional wind (contour), averaged between 60° and 80°E in May. There is a warming center around the Tibetan Plateau. The meridional circulation is southward winds (positive means south wind) in uppers and northward winds in lows.
https://static-content.springer.com/image/art%3A10.1007%2Fs00382-010-0982-0/MediaObjects/382_2010_982_Fig12_HTML.gif
Fig. 12

a The differences (SEN1-CON) of latitude-height cross-section for mean temperature (shade, units: °C) and meridional wind (contour, units: m/s) from 60° to 80°E, b averaged from 80° to 100°E

Figure 12b is the same as Fig. 12a, but averaged between 80° and 100°E. Lower troposphere between 15°N and 30°N is heated and air temperature rises via dry convection along the southern slope of Tibetan Plateau, producing a positive temperature anomaly in the mid-to-upper troposphere. Warming air rising forced by the increasing heat leads to upward motion of air which causes low-level air pressure decreasing, and then the north–south pressure gradient is amplified. It is so called “elevated heat pump” by Lau and Kim (2006). Although, between 60° and 80°E, there is also a rising temperature center in mid-to-upper troposphere, but the warming process is not clearer than this region (80–100°E).

As a result, there is a meridional circulation with northward winds in uppers and southward winds in lows. The warm and moist air from Indian Ocean is transporting to northeast India and Myanmar and enhances precipitation over this region. Finally, the increased precipitation sets the stage for the advanced onset of the Indian monsoon. Lau et al. (2006) found that the increased dust loading coupled with black carbon emission from local sources in the northern India during late spring may lead to an advance of the rainy periods and subsequently an intensification of the Indian summer monsoon. Zhang et al. (2009) note that the carbonaceous aerosols (BC, OC) weak the Hadley circulations in the Northern Hemisphere and increasing precipitation. Though our experiment includes three types of aerosols (sulfate, BC and OC), we also find “elevated heat pump” and the similar conclusion with formers.

4 Conclusion

We use a coupled regional climate-chemistry model to conduct sensitivity experiments about anthropogenic aerosols (sulfate, BC, OC) and their climatic effects on Indian monsoon. First, we evaluate the model performance against available observations. Then, we analyze the aerosols’ impact on the Indian summer monsoon.

The results show that the model represents the basic climatology over the region and the chemistry model performs relatively well in simulating AOD and near surface concentration. The primary model shortages are an underestimate AOD and near surface concentration over South Asia. Model errors in simulated climatic patterns are main reasons about the deviation between simulations and observations. (e.g. There are errors of simulated regional climatic winds and precipitation which would strongly affect the aerosols.). The coupled chemistry module also has limitations and should be mended in future. Our experiment does not include the other aerosols (e.g. Dust aerosols), these lacks directly lead AOD underestimated.

Negative short wave radiative forcing is found at both the top of atmosphere and the surface. The magnitude of the forcing is in good correspondence to the AOD and aerosols’ concentration. The greatest radiative forcing is found in northeast India both in pre-monsoon and monsoon season.

In pre-monsoon season (SEN1), aerosols result in surface air temperature decreasing by 0.1–0.5°C over the northeast India and Myanmar. Meanwhile, the precipitation increases by 5–25% in these regions. These changes are more consistent with monsoon onset variety. In the areas of increasing precipitation, monsoon breaks out earlier and vice versa.

In monsoon season (SEN1), the surface air temperature cools by 0.1–0.5°C in northwest and northeast Indian subcontinent and accompanies with increasing precipitation by 5–25%. In addition, we know that Tibetan Plateau is very clean, nearly zero emission (Fig. 1), and the change of surface air temperature is little in pre-monsoon and monsoon seasons, but the impact on precipitation is clear in the southern Tibetan Plateau.

The temperature decreases both in SEN2 and SEN3. Cooling extent and intensity are largely depended on aerosols’ concentration. Changes in precipitation do not directly correspond with aerosol distribution. The aerosols’ radiation forcing impact on thermal balance of mid-level atmosphere, induce atmospheric circulation anomalies and indirectly change precipitation.

The aerosols lead to monsoon onset date advancing by 1–2 pentads in Myanmar and northeast India (SEN1), since the aerosols weaken the local Hadley circulations and increase the precipitation. The result is consistent with the Lau et al. (2006) and Zhang et al. (2009). Little changes are found in the west India with half emissions. When emission doubled, different pattern are existed form SEN1 and SEN2 over the India subcontinent. However, the monsoon onset dates are yet ahead by 1–2 pentads in the region of Myanmar and northeast India where the rainfall increased.

It should be noted that the simulation of aerosols and their climatic effects are still very complicated. There are many uncertainties, especially in the simulation of the impact on temperature and precipitation. Our study is limited in two aspects. Firstly, we do not include the contribution of dust aerosols which have been proven to be important over South Asia (Lau et al. 2006). This is because our major focus is to investigate effects of anthropogenic aerosols on climate. Our modeling system has the capability of including dust aerosols (Zakey et al. 2006; Zhang et al. 2009), thus our further work is to perform new sets of simulations including all type of aerosol. Secondly, the emission inventory does not include seasonal transformation. Actually, the black carbon has an evident seasonal diversification (Habib et al. 2006).

The model and its chemistry module need to be improved in near future, such as induct indirect effect, e.g. the role of particles as cloud condensation nuclei, are not described in our simulation. Eventually we plan to use the full model with multiple aerosols and direct and indirect effects for a new set of climate change simulations over South Asia.

Acknowledgments

This study is supported by the National Basic Research Program of China (2009CB421407, 2010CB950102), the Chinese Academy of Sciences (KZCX2-YW-145), the National Nature Science Foundation of China (40830743, 40771187), State Key Laboratory of Cryospheric Science (SKLCS-ZZ-2008-01).

Copyright information

© Springer-Verlag 2010