Temporal characteristics of aerosol optical properties over the glacier region of northern Pakistan

Abstract

Glacier melting due to light-absorbing aerosol has become a growing issue in recent decades. The emphasis of this study is to examine aerosol loadings over the high mountain glacier region of northern Pakistan between 2004 and 2016, with sources including local emissions and long-range transported pollution. Optical properties of aerosols were seasonally analyzed over the glacier region (35–36.5°N; 74.5–77.5°E) along with three selected sites (Gilgit, Skardu, and Diamar) based on the Ozone Monitoring Instrument (OMI). The aerosol sub-type profile was analyzed with Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO). Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model was used to understand the origin of air masses arriving in the study region. The highest values of aerosol optical depth (AOD) and single scattering albedo (SSA) occurred during spring, whereas aerosol index (AI) and absorption AOD (AAOD) exhibited maximum values in winter and summer, respectively. The minimum values of AOD, AI, AAOD, and SSA occurred in winter, autumn, winter, and autumn, respectively. The results revealed that in spring and summer the prominent aerosols were dust, whereas, in autumn and winter, anthropogenic aerosols were prominent. Trend analysis showed that AI, AOD, and AAOD increased at the rate of 0.005, 0.006, and 0.0001 yr⁻¹, respectively, while SSA decreased at the rate of 0.0002 yr⁻¹. This is suggestive of the enhancement in aerosol types over the region with time that accelerates melting of ice. CALIPSO data indicate that the regional aerosol was mostly comprised of sub-types categorized as dust, polluted dust, smoke, and clean continental. The types of aerosols defined by OMI were in good agreement with CALIPSO retrievals. Analysis of the National Oceanic and Atmospheric Administration Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model revealed that air parcels arriving at the glacier region stemmed from different source sites.

1. Introduction

The earth’s environment is continuously deteriorating due to increasing human population and corresponding activities leading to pollutant emissions. For example, industrial activities, land use, and combustion of fossil fuels emit greenhouse gases and aerosols, resulting in atmospheric composition changes (Mann et al., 1999; Crowley, 2000; Houghton et al., 2001). Aerosols can impact air quality, public health, and climate via both direct and indirect effects. The direct effect on climate occurs when particles scatter and absorb solar radiation, whereas the indirect effect is brought about by how more aerosol particles lead to more numerous but smaller droplets, yielding more reflective clouds at fixed cloud liquid water path (Twomey, 1977).

Scattering of solar radiation by aerosols and clouds causes negative forcing, which tends to cool the surface of the earth, whereas the absorption of terrestrial radiations by aerosols and clouds cause a positive forcing, and thus warming (Hinds, 1999). These effects are quantified in the form of Aerosol Radiative Forcing (ARF). Information needed to robustly calculate ARF includes the concentration of aerosols and its other characteristics, such as composition, size, and optical properties (Russell et al., 2010). Changes in aerosol optical properties arise owing to different emissions sources, chemical transformations, and long range transport of pollution (Ram et al., 2010).

The scattering and absorption of light by aerosols depends on composition and size. Aerosols like inorganic salts (e.g., ammonium sulfate and ammonium nitrate) and sea salt scatter solar radiation back to space resulting in cooling of the planet. However, black carbon (BC) and iron oxides absorb radiation, which leads to heating (Wang et al., 2011). Deposition of BC aerosols on snow/ice surfaces has accelerated snow melt and glacier retreat over the Himalaya and Tibetan Plateau (Zhang et al., 2015). A key source of BC is biomass burning in the region (Rasul et al., 2011). Mineral dust changes the albedo of glacier surfaces, which in turn affects the energy balance of the atmosphere as well as the melting rate of glaciers and seasonal snow (Fujita, 2007). There are different sources of dust deposited on glaciers, including transported desert dust and locally produced mineral dust (Shahgedanova et al., 2013). Analysis of the sensitivity of Vadret da Morteratsch Glaciers showed that a 10% decrease in reflectance from a glacier surface yielded an equivalent impact of 1.7 K warming (Oerlemans et al., 2009); therefore, changes in deposition of absorbing aerosol types can have a major impact on radiative forcing.

Pakistan is known to have the world’s highest mountains. Three world famous mountain ranges, specifically the Himalaya, Karakorum, and Hindukush, join in the extreme northern part of Pakistan, which is occupied by glaciers. There are > 5000 glaciers in the Pakistani geographical area feeding the Indus River. The frozen water resources are decreasing continuously due to global warming effects, reducing ice mass (Rasul et al., 20011). The increasing trend in temperature has been observed during the last decade in the northern part of Pakistan, which enhances the snow/ice melt (Rasul et al., 2011), affecting agriculture, drinking water supplies, and hydroelectric power, amongst other sociologically-relevant necessities.

In Pakistan, many studies have been carried out on aerosol optical properties over plain areas (Alam et al., 2011, 2012, 2014a,b; Bibi et al., 2015; Tariq and Ali, 2015; Bibi et al., 2017a,b; Iftikhar et al., 2018; Zeb et al., 2018). But no study to our knowledge has been conducted on aerosol optical properties over the glacier areas in northern Pakistan. The aim of this study is to fill in that gap by reporting a comprehensive characterization of aerosol optical properties on both local and regional scales. The scientific gap in our present knowledge about the forcing implications over the glacier region is crucial for climate modelling tasks, and adaptation to the potential effects of changing ice melt rates.

2. Study area and datasets

2.1. Site description and local meteorological conditions

The present study was conducted over a glacier area located in the extreme north of Pakistan (35–36.5°N; 74.5–77.5°E; see Fig. 1). In addition to study this on a regional scale, individual locations are examined in the province of Gilgit-Baltistan, including Gilgit (35.9⁰N, 74.3⁰E), Skardu (35.29⁰N; 75.62⁰E), and Diamar (35.43⁰N; 73.93⁰E). The glacier areas in Pakistan are spread over an area of about 16933 km² and consist of high altitude peaks and lakes. The three well-known mountain ranges (i.e., Himalayas, Karakoram, and Hindu Kush (HKH)) are joined together in this area. Daily data over the study period (2004–2016) were classified into four seasons: spring (March to May), summer (June to August), autumn (September to November), and winter (December to February).

Fig. 1.

Map of the study area in Pakistan.

Meteorological data sets were obtained from the Pakistan Meteorological Department in Gilgit-Baltistan. The meteorological situation in Gilgit, Skardu, and Diamar are summarized in Fig. 2a. In Gilgit, Skardu, and Diamar maximum precipitation accumulation values of 18.94, 31.24 and 25.59 mm were recorded, respectively, during spring months. In Gilgit, the minimum precipitation of 7.44 mm was observed during winter, while Skardu and Diamar exhibited minimum precipitation values of 9.64 and 6.10 mm, respectively, during the autumn months. Maximum relative humidity (RH) values at Diamar and Skardu were 56.38% and 81.59%, respectively, during winter, while in Gilgit it was 77.60% both in winter and autumn. Maximum temperatures were observed during the summer with values of 34.93 °C (Gilgit), 30.42 °C (Skardu), and 37.64 °C (Diamar). Wind speed varied from 0.1 to 2.1 ms⁻¹, 0.1 to 2.4 ms⁻¹, and 0.1 to 3.9 ms⁻¹ in Gilgit, Skardu, and Diamer, respectively, with corresponding averaged wind speeds of 0.54, 0.61, and 1.29 m s⁻¹ (Fig. 2b–d). In Gilgit and Diamer, the winds were predominantly westerly, while in Skardu the winds were mainly southeasterly and northwesterly.

Fig. 2.

Prevailing meteorological conditions of Gilgit, Skardu, and Diamar from 2004 to 2016. Panels b, c, and d correspond to Gilgit, Skardu, and Diamer, respectively.

2.2. Datasets and tool used

2.2.1. Ozone Monitoring Instrument (OMI)

The OMI aboard the Aura satellite was launched in July 2004 and is nadir-viewing with a spatial resolution (13–24 km) that measures the Top of Atmosphere (TOA) upwelling radiances in the solar spectrum range between 270 and 500 nm (Levelt et al., 2006). OMI is able to detect layers of absorbing aerosols produced from various sources like biomass burning and desert dust plumes. Torres et al. (2007) provided a detailed method of obtaining the Ultraviolet (UV) AI using the “near UV Aerosol Retrievals (OMAERUV)” algorithm, which can be used for the identification of some prominent absorbing aerosols like desert dust and carbonaceous aerosols. For this study, daily Level 3 AI (1° × 1°) OMT03d (version 003) data are used between 2004 and 2016. The AOD, AAOD, and SSA daily product at 500 nm with spatial resolution of 0.25° × 0.25° was also used. Further information regarding the details of data sets is available on the OMI website (https://aura.gsfc.nasa.gov/omi.html). In the current study, OMI data are utilized for the seasonal analysis of Aerosol Optical Depth (AOD), Aerosol Index (AI), Absorption Aerosol Optical Depth (AAOD), and Single Scattering Albedo (SSA). OMAERUV AOD values exhibit an uncertainty maximum of 30% at 440 nm. According to the results provided by Bais et al. (2005), the uncertainty of SSA retrievals derived by global UV irradiances with overall accuracy (derived from the calibration) on the order of 5% can vary from ± 0.05 (for high AOD and SSA values) to ± 0.15 (for low AOD and SSA values). Assuming the error in AAOD ≈ δSSA· AOD (Arola et al., 2005), the uncertainty in estimated AAOD values can vary from 0.020 (high aerosol load) to 0.026 (low aerosol load).

2.2.2. Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO)

The CALIPSO satellite was launched on 28 April 2006 and provides information about the radiative effects of aerosols. CALIPSO carries the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument, which is functional on two wavelengths (532 nm and 1064 nm) and provides measurements of attenuated backscatter radiation around the entire globe. Key retrieved parameters include total attenuated backscatter at 532 nm and 1064 nm, their attenuated color ratio (1064/532 nm), perpendicular backscatter at 532 nm, and the depolarization ratio (DR). The CALIOP instrument can provide the unique vertical profile of both aerosols and clouds by using an active laser beam (Mishchenko et al., 1999; Winker et al., 2003). The equatorial crossing times of CALIPSO are about 13:30 and 01:30, with a repeating cycle of 16 days (Winker et al., 2009). CALIPSO can detect aerosols over bright surfaces, in clear sky conditions as well as beneath thin clouds (Bibi et al., 2015, Huang et al., 2007a,b; Winker et al., 2007; Liu et al., 2008a,b; Geng et al., 2011). The CALIPSO product is highly useful in discriminating aerosol layers from clouds (Liu et al., 2009) and categorizing aerosol layers as one of six subtypes (dust, marine, smoke, polluted dust, polluted continental, and clean continental) (Omar et al., 2009).

Volume depolarization ratio (VDR) is an important observation from the CALIOP sensor (Winker et al., 2007), which is defined as the ratio of the parallel and perpendicular components of incoming lidar signals at 532 nm:

$$\mathit{\text{VDR}}(z)=\frac{{\beta }_{532,\perp }^{\prime }(z)}{{\beta }_{532,\Vert }^{\prime }(z)}$$ (1)

where z is height, while ${\beta }_{532,\perp }^{\prime }\left(z\right)$ and ${\beta }_{532,\Vert }^{\prime }\left(z\right)$ are the perpendicular and parallel components of the attenuated backscatter, respectively.

VDR quantifies the degree of sphericity of particles, with larger values (≥0.5) indicating irregular particles such as dust, and lower values (~0.1–0.2) indicating spherical particles such as BC, sulfate, and products of biomass burning (Liu et al., 2008b; Omar et al., 2009). In this work, CALIPSO data are used for DR and detection of key aerosol sub-types impacting the study region.

2.3. Methods

2.3.1. Aerosol optical properties

In order to understand the aerosol optical properties over glacier region, we have examined AOD, AI, AAOD, and SSA, obtained from OMI instrument. Aerosol Optical Depth (AOD) is a parameter quantifying total columnar extinction (sum of scattering and absorption), which is related to the extent to which aerosols can directly influence the radiative balance at the surface and at the TOA.

$$\text{AOD}(\lambda )=\underset{{\text{h}}_1}{\overset{{\text{h}}_2}{\int }}{\beta }_{\text{ext},\lambda }(\text{h})\text{dh}$$ (2)

where λ and β_ext represent the wavelength and extinction coefficient of aerosols, respectively, and h_1, and h₂ are the lower and upper altitude in units of length.

Ultraviolet Aerosol Index (UVAI), referred to hereafter as AI, is an index to distinguish between absorbing and non-absorbing aerosols in the UV range. The AI is basically a quantification of the change in the quantity of Rayleigh scattered light measured at the top of the atmosphere caused by the existence of scattering and absorbing particles (Herman et al., 1997; Torres et al., 1998). AI is actually based on a spectral contrast method in a UV region where ozone absorption is very small. Backscattered radiation at wavelengths of 340, 360 and 380 nm is caused mainly by molecular Rayleigh scattering, terrestrial reflection and diffusion by aerosols and clouds (through Mie scattering). It is the difference between the observations and model calculations of absorbing and non-absorbing spectral radiances. AI can characterize the absorbing types of aerosol (e.g., dust and biomass burning) (De Graaf et al., 2005; Buchard et al., 2015).

AI can be expressed as

$$\mathit{\text{AI}}=100{\mathit{\text{log}}}_{10}\left[I_{360-\mathit{\text{means}}}/I_{360-\mathit{\text{calc}}}\right]$$ (3)

where I_360-means, is the backscattered radiance at the wavelength measured by TOMS and OMI (with Mie and Rayleigh scattering, and absorption) and I_360-calc is the radiance calculated from the model molecular atmosphere with Rayleigh scatterers. I₃₆₀ depends strongly on the absorbing optical thickness of the Mie scatterers. AI exhibits a positive value for UV-absorbing aerosols (e.g., mineral dust, soot, and smoke particles) and a negative value for non-absorbing aerosols (e.g., sulfate aerosols and sea salt particles) of both natural and anthropogenic origin. Clouds, on the other hand, exhibit AI values near zero (Torres et al., 1998).

Absorption Aerosols Optical Depth (AAOD) is a columnar estimate of the concentration of near-UV absorbing aerosol particles (i.e., smoke and mineral dust) and is retrieved using OMI (Bucsela et al., 2008). AAOD can be obtained by using the following equation if the value of SSA and AOD are known:

$$\mathit{\text{AAOD}}(\lambda )=[1-\mathit{\text{SSA\hspace{0.17em}}}(\lambda )]*\mathit{\text{\hspace{0.17em}AOD\hspace{0.17em}}}(\lambda )$$ (4)

where λ is the wavelength. We use the AAOD daily product at 500 nm (OMAERUVd.003).

Single Scattering Albedo (SSA) is a key parameter providing quantitative details about the absorbing and scattering nature of aerosols, and can be used to derive direct aerosol radiative forcing. Scattering and absorbing characteristics of aerosols, together with surface reflectance, determine whether aerosols contribute towards cooling or heating of the atmosphere (Lewis et al., 2008). SSA is usually expressed by the following equation:

$$\text{SSA}=\frac{\text{Qscat}}{\text{Qext}}=\frac{\text{Qscat}}{\text{Qscat\hspace{0.17em}}+\text{\hspace{0.17em}Qabs}}$$ (5)

where Q_sca represent scattering efficiency and Q_ext denote extinction efficiency, which is the sum of absorption and scattering efficiencies i. e (Q_ext = Q_abs + Q_sca). The value of SSA range from 0 (complete absorption) to 1 for purely scattering types of aerosols. The value of SSA depends on the value of aerosol size, composition, concentration of absorbing components and their degree of mixing with non-absorbing components.

3. Results and discussion

3.1. Variations in aerosol optical properties

3.1.1. Aerosol optical depth

In this study, the seasonal and monthly variations in AOD values were analyzed with OMI data at 500 nm for the period between 2004 and 2016. Fig. 3a–b showed the seasonal and monthly variations in AOD over the study region. The results reveal that AOD values were higher during spring, followed by summer, autumn, and winter. During spring and summer, the AOD values ranged from 0.49 to 0.67 and 0.49 to 0.65, respectively. Similarly, the AOD value for autumn and winter varied between 0.27 to 0.43 and 0.22 to 0.33, respectively. For the whole area, the highest monthly mean AOD value of 0.65 ± 0.11 was observed during May and the lowest value of 0.24 ± 0.04 was noted during December.

Fig. 3.

Box-whisker plots summarizing (a) seasonal variations in AOD over the entire study region, and (b) monthly variation in AOD over the entire region. Each box indicates the 25th and 75th percentiles and the whiskers show the 5th and 95th percentiles. The vertical lines show the standard deviation from the mean value. The solid circle inside each box and the horizontal lines represent the mean and median values, respectively. The crosses above and below the boxes indicate the maximum and minimum values, respectively.

Table 1 summarizes how AOD characteristics vary in all three locations of the Gilgit-Baltistan province relative to the entire study region for each season. Similarly with the cumulative region, each of the three sites, exhibits the maximum AOD during spring, followed by the summer, autumn, and winter. The maximum value of AOD in spring is largely due to vehicular emissions, biomass burning, local pollutants, and strong wind erosion leading to dust over glacier region. High AOD during spring is also due to long-range transport of dust brought by westerly winds from different regions like the Sahara Desert, West Asia, the Thar Desert, and the Middle East; those air masses pass over the Indo-Gangetic plain and reach the Himalayan region (Kayetha et al., 2007). During the pre-monsoon (spring) season, because of the higher convection and abrupt pressure gradient, dust-rich aerosols reach high elevated stations along the slopes of the Himalayas (Gautam et al., 2009). The significant AOD increase at the high altitude site (Manora Peak) in central Himalaya is due to the increase in concentration of mineral dust, which is linked with long-range transport from desert regions (Ram et al., 2010). The enhanced fire activities in the pre-monsoon (spring) increases the AOD value over North-East India (Kharol et al., 2008). Chatterjee et al. (2012) also reported maximum values of AOD in spring over Darjeeling, Eastern Himalaya. They also found high AOD (0.50) during May and low AOD (0.23) during August (2012) over Darjeeling, eastern Himalayan. The reason for higher AOD over Darleging in the spring season is due to dry lands because the increase in the solar heating rate causes dust to rise from arid and semi-arid regions of Western India and from West Asia. Dust is the main contributor to the increased AOD during spring and summer as compared to the other types of aerosols over the Tibetan plateau (Xu et al., 2015). Higher values of AOD in spring and summer are because of large concentration of water vapor, also due to high temperature and high wind speeds that promotes soil dust emissions (Alam et al., 2011). Lower values of AOD observed in winter and autumn are due to reduced surface emissions of primary aerosols and washout processes that occur due to heavy rains in most sites over glacier region, which removes aerosols from the environment (Papdimas et al., 2008; Crosbie et al., 2014; Alam et al., 2014a). Ningombam et al. (2014) found the AOD values of 0.044 ± 0.002, 0.031 ± 0.001, 0.031 ± 0.001, and 0.061 ± 0.002, during summer, autumn, winter, and spring, respectively at 500 nm wavelength over Hanle (high-altitude location), in the trans-Himalayan region. Guleria et al. (2012) investigated the AOD during April 2006–March 2009 over Mohal in the Kullu Valley (North Western Himalaya of India). They found the AOD value of 0.24 ± 0.08 at 500 nm for the entire study period. The maximum AOD of 0.34 ± 0.08 was found during pre-monsoon, followed by 0.26 ± 0.08, 0.21 ± 0.05 and 0.20 ± 0.07 during monsoon, post-monsoon, and winter, respectively. Kant et al. (2015) noted the AOD value of 0.62 ± 0.11 and 0.56 ± 0.21 during premonsoon season of 2013 over two high altitude locations location i.e. Dehradun and Patiala.

Table 1.

Seasonal average variation in AOD, AI, AAOD, and SSA with the corresponding standard deviation (STD) over the study area for the period between 2004 and 2016.

Region	Seasons	AOD ± STD	AI ± STD	AAOD ± STD	SSA ± STD
Gilgit	Spring	0.53 ± 0.05	0.78 ± 0.19	0.01 ± 0.005	0.98 ± 0.07
	Summer	0.45 ± 0.03	0.77 ± 0.23	0.02 ± 0.002	0.96 ± 0.04
	Autumn	0.31 ± 0.06	0.75 ± 0.21	0.01 ± 0.004	0.96 ± 0.06
	Winter	0.32 ± 0.07	0.81 ± 0.34	0.01 ± 0.003	0.96 ± 0.03
Skardu	Spring	0.66 ± 0.15	0.75 ± 0.06	0.01 ± 0.07	0.98 ± 0.01
	Summer	0.50 ± 0.06	0.78 ± 0.07	0.02 ± 0.003	0.96 ± 0.01
	Autumn	0.38 ± 0.07	0.85 ± 0.62	0.02 ± 0.005	0.96 ± 0.08
	Winter	0.32 ± 0.01	0.81 ± 0.53	0.006 ± 0.002	0.97 ± 0.01
Diamer	Spring	0.46 ± 0.06	0.76 ± 0.15	0.02 ± 0.004	0.96 ± 0.01
	Summer	0.44 ± 0.05	0.68 ± 0.1	0.02 ± 0.003	0.96 ± 0.005
	Autumn	0.27 ± 0.03	0.80 ± 0.34	0.01 ± 0.003	0.95 ± 0.01
	Winter	0.26 ± 0.04	0.95 ± 0.54	0.01 ± 0.005	0.96 ± 0.01
Whole area	Spring	0.60 ± 0.06	0.72 ± 0.06	0.017 ± 0.004	0.97 ± 0.06
	Summer	0.57 ± 0.06	0.71 ± 0.04	0.020 ± 0.004	0.96 ± 0.003
	Autumn	0.34 ± 0.05	.68 ± 0.09	0.015 ± 0.003	0.95 ± 0.005
	Winter	0.28 ± 0.04	0.78 ± 0.2	0.009 ± 0.004	0.97 ± 0.01

He et al. (2017) studied the optical properties of aerosol and their corresponding radiative forcing over the Yangtze River Basin from 2001 to 2015. They observed high AOD (0.58 ± 0.35) during spring and low AOD (0.42 ± 0.29) during the winter season with an average value of 0.49 ± 0.31. Wang et al. (2015) conducted the long term study (2007–2013) of the optical properties of aerosol i. e AOD, angstrom exponent, SSA, refractive index, and size distribution of aerosol over Wuhan (urban location in Central China). They found relatively high values of AOD_{500 nm} throughout the year with maximum value of 1.52 in June 2012 and minimum value of 0.57 in November 2012. Zhang et al. (2018a,b) studied the spatio-temporal variation of aerosol over the Yangtze River Basin from 2001 to 2015 using MODIS, multi-angle imaging spectroradiometer (MISR), and ground-level particulate matter (PM) data. The observed the AOD values greater than 0.8 over Yangtze River Delta, central China, and the Sichuan Basin, while the AOD values less than 0.3 were found over higher-elevation areas in the western Sichuan Plateau and the source regions of the Yangtze River. Zhang et al. (2018a,b) studied the AOD variations and radiative properties of aerosols over South China and its adjacent area from 2001 to 2016 using MODIS and OMI satellite data. They observed high AOD (0.7) during spring and low AOD (0.4) during the winter season. Zhang et al. (2017) also studied the aerosol optical and radiative properties over central China. They found high AOD values over northern and central parts, whereas low AOD values were noted over the southern and western parts of China. Bibi et al. (2015) conducted the inter-comparison of aerosol optical depth (AOD) retrievals from multiple satellite sensors, such as MODIS, MISR, OMI, and CALIPSO over four locations in the Indo-Gangetic plain during 2007–2013 and found high AOD values during the summer and low during winter.

3.1.2. Ultraviolet Aerosol Index

Seasonal and monthly variations in AI values over the whole region are shown in Fig. 4a–b. The highest value of AI was observed in winter and the lowest value was found in autumn over the whole region. During winter, spring, summer, and autumn, the AI values ranged from 0.60 to 1.28, 0.64–0.87, 0.64–0.78, and 0.58–0.90, respectively. These values of AI were due to aerosol (dust, polluted dust and smoke) loading over the region. For the whole area, the highest and lowest monthly mean AI values of 0.86 ± 0.34 and 0.64 ± 0.07, respectively, were observed during January and October. Table 1 shows the seasonal average values of AI for the study region. No negative value of AI was obtained over the study area, which indicates the dominance of UV-absorbing aerosols over these regions.

Fig. 4.

Box-whisker plots indicating (a) seasonal variations in AI over the entire study region, and (b) monthly variation in AI over the entire study region. The box-whisker representation in all panels is the same as in Fig. 3.

Seasonal variations in AI were associated with the seasonal variations in dust, smoke, rainfall, and burning activities in the three regions. During the winter season, the weather in the upper regions of the northern area is very cold. The maximum absorbing aerosols in the winter season are mostly fine particles (Bibi et al., 2017a,b), which occur due to burning of woods, dung, and other fuels (Prospero et al., 2002). During the autumn months, increase in temperature, wind speed, and dust storm activities promote enhanced AI values (Tariq and Ali, 2015). In summer, due to the unstable atmospheric conditions, aerosols are lifted to greater heights, which results in less AI values as compared to other seasons. (Tariq and Ali, 2015; Bibi et al., 2017a,b). Chatterjee et al. (2012) reported that high values of AI during spring over the Himalaya stem from windblown dust driven by westerlies.

3.1.3. Absorbing aerosol optical depth

The seasonal and monthly variation in AAOD over the whole region is shown in Fig. 5a–b. AAOD was observed to be a maximum during summer and a minimum during winter. During spring, summer, autumn, and winter, AAOD values ranged from 0.013 to 0.028, 0.017–0.024, 0.010–0.017, and 0.005–0.019, respectively. For the whole area, the highest monthly mean AAOD value of 0.02 ± 0.007 was observed during April and the lowest value of 0.006 ± 0.005 was observed during February.

Fig. 5.

Box-whisker plots indicating (a) seasonal variations in AAOD over the entire study region, and (b) monthly variation in AAOD over the entire study region. The box-whisker representation in all panels is the same as in Fig. 3.

AAOD is one of the absorbing properties of aerosols, where BC and dust are the prominent contributors (Bibi et al., 2017a,b), consistent with the sources impacting the study region. Vadrevu et al. (2015) reported AAOD values of 0.075 and 0.05 during summer and winter, respectively, over Punjab (India). Bibi et al., 2017a,b reported AAOD values in the range of 0.01–0.065 over Karachi. Koike et al. (2014) observed that high dust loading is responsible for high value of AAOD in South Asia based on in situ measurements. Kang et al. (2017) conducted a study of absorbing aerosols from 2005 to 2016 and found that the AAOD values range from 0.01 to 0.04 over different locations over East Asia during four seasons.

3.1.4. Single scattering albedo

Fig. 6a–b shows the seasonal and monthly variation in SSA for the whole region. Values of SSA ranged between 0.96 and 0.98, 0.96–0.97, 0.95–0.96 and 0.94–0.98 during spring, summer, autumn, and winter, respectively. The maximum average SSA was observed in February (0.98 ± 0.004) and the minimum SSA was observed in October (0.96 ± 0.003) over the whole region. Average values of SSA in Gilgit, Skardu, and Diamar are summarized in Table 1.

Fig. 6.

Box-whisker plots indicating (a) seasonal variations in SSA over the whole region, and (b) monthly variations in SSA over whole region. The box-whisker representation in all panels is the same as in Fig. 3.

Overall, it was noted that SSA values were mostly high during winter because of the existence of a diverse combination of anthropogenic aerosols (fossil fuel combustion, organic aerosols and absorbing soot), while during summer and spring, SSA levels were driven more by coarse particles (e.g., dust), long-range transport, and high humidity, which could potentially promote production of more water-soluble aerosol. During pre-monsoon (spring) and monsoon seasons, enhanced water vapor levels are present in the atmosphere and particles swell owing to both hygroscopic growth and heterogeneous chemistry (e.g., Sorooshian et al., 2011; Youn et al., 2013), resulting in greater SSA values at higher wavelengths (Singh et al., 2004). Desert dust transport events over India and China can significantly change SSA values ranging from 0.75 to 0.99 (Redmond et al., 2010). On the other hand, during autumn, SSA values are lower as compared to other seasons, indicating that dust is not a major contributor but local fine pollution aerosols are dominant.

Pant et al. (2006) reported that the SSA values over Monora Peak ranged between 0.87 and 0.94, with an average value of 0.90 during winter. Over the Himalayan region in Nepal, Ramana et al. (2004) found that SSA ranged between 0.7 and 0.9 in the winter of 2004. Gautam et al. (2011) reported that during the pre-monsoon season, the SSA values along with standard deviations at different sites along the Himalayan foothills (i.e.,Hetauda, Dhulikhel, Langtang) were 0.86 ± 0.02, 0.88 ± 0.02, and 0.89 ± 0.03, respectively, at a wavelength of 441 nm. Marcq et al. (2010) showed that SSA values varied between 0.82 and 0.89 at the Nepal Climate Observatory-Pyramid during the spring season. Nair et al. (2013) noted SSA values of 0.97 and 0.96 over Himalayan stations like Hanle during pre-monsoon and winter seasons. Verma et al. (2010) found the yearly average value of SSA (0.96) at 500 nm over Hanle (4520 m asl, Western Himalaya) during 2007. Kang et al. (2017) studied the absorbing aerosols over East Asia from 2005 to 2016 and found that SSA range from 0.94 to 0.98. Kant et al. (2015) carried out study over two locations, i.e. Dehradun and Patiala (North-west, India) during premonsoon 2013. They found the lower value of SSA (500 nm) over Patiala (0.83 ± 0.01) as compared to Dehradun (0.90 ± 0.01), which suggest the high amount of absorbing of aerosols over Patiala.

3.1.5. Frequency distribution of seasonal AOD, AI, AAOD, and SSA

In Fig. 7, variations in AOD, AI, AAOD, and SSA are summarized using a histogram format. In each plot, the number of observed days and seasonal average values with associated standard deviations are given for each parameter. For each season, a large difference in the distributions for each parameter was noted, signifying the considerable seasonal heterogeneity due to differences in both emission sources and meteorological conditions (Tiwari et al., 2016). Frequency distribution analysis shows that AOD values greater than 0.6 accounted for 75% of data during the summer, 68% during spring, 14% during autumn, and 8% during winter suggestive of high aerosol loading during summer and spring. The frequency of AI > 0.5 was 100% during all the four seasons, confirming the existence of absorbing aerosols in all seasons. AAOD frequency was observed to be highest in summer (84%) followed by autumn (69%), spring (68%), and winter (42%), the maximum value during summer is due to high absorption of coarse particles at lower wavelengths. Likewise, the SSA values are in the range from 0.90 to 1, suggest the dominance of scattering aerosol. The frequency distributions of SSA (greater than 0.96) during winter, spring, autumn and summer were found to be 69%, 61%, 42%, and 37%, respectively. Recently, Tiwari et al. (2016) also reported the same type of frequency distribution for New Delhi, India, they further noted that during the four seasons, the large variations in AOD > 0.7 reveal different particle types produced by diverse emission sources that show a strong seasonal effect.

Fig. 7.

Seasonal frequency distributions (%) (a) AOD, (b) AI, (c) AAOD, and (d) SSA with N and m corresponding to the number of observed days and the seasonal mean, respectively, over the study region between 2004 and 2016.

3.1.6. Trend analysis of seasonal AOD, AI, AAOD, and SSA

Inter-annual variations in optical parameters (AOD, AI, AAOD, SSA) are summarized in Fig. 8a–d over the study region between 2004 and 2016. To calculate the slope, intercept, and trend, a linear regression method is used (Wilks, 2006). AOD exhibits an increasing tendency (0.006 yr⁻¹), with the lowest and highest values of 0.21 and 0.61 observed in winter 2010 and summer 2016, respectively, with an overall average of 0.44 ± 0.15. An increasing trend for AI (0.005 yr⁻¹) was noted over the study region (Fig. 8b). The lowest seasonal AI value of 0.57 was observed in autumn 2004 and the highest seasonal AI value of 1.27 was observed in winter 2016 with an overall average value of 0.72 ± 0.12. An increasing trend of AAOD (0.0001 yr⁻¹) was also observed over the study region (Fig. 8c). The minimum AAOD value of 0.005 was observed in winter 2007 and the maximum value of 0.28 was observed in spring 2014, with an overall average of 0.015 ± 0.004. A decreasing trend of SSA (−0.0002 yr⁻¹) was noted over the study region (Fig. 8d). The SSA values of 0.94 and 0.98 were observed in autumn 2016 and winter 2004, respectively, with an average value of 0.96 ± 0.12. These values indicate the dominance of scattering aerosol over the absorbing aerosol. All the trend results show the increasing amount of both scattering and absorbing aerosols as a function of time, although the types of these aerosols differ seasonally.

Fig. 8.

Time series of (a) AOD, (b) AI, (c) AAOD, and (d) SSA over the study site for the period between 2004 and 2016.

Zhang et al. (2018) found an increasing trend of AOD from 2001 to 2004 and a decreasing trend from 2004 to 2016 over South China and its adjacent areas. They attributed the change due to pollutant discharging. Zhang et al. (2017) also investigated the significant decreasing trend of AOD with a maximum decrease rate (−0.08 per year) in the northern and western parts, which were attributed to the decreasing emission of aerosols and increasing amount of rainfall. He et al. (2018) analyzed the trends in AOD and aerosol direct radiative effect (ADRE) in clear sky over the Yangtze River Basin (YRB) from 1980 to 2016 using the MERRA-2 product. They observed the increasing trend (0.0201 year⁻¹) of AOD up to 2008 and then decreasing trend (−0.0185 year⁻¹) up to 2016. They attributed the decreasing trend due to the decreases of S0₂, NO_x and soot emissions from major cities in the YRB since 2006. Alam et al. (2011) noted the increasing tendency for AI over ten selected locations in Pakistan. Similarly, Tariq and Ali, 2015 also pointed out the increasing trend for AI over Pakistan.

This increasing trend in aerosol, reduce the incoming solar flux both at the earth’s surface and top of the atmosphere, consequently, heat up within the atmosphere. Furthermore, the deposition of these aerosol particles on snow/ice surfaces enhance glacier melt. Consequently, a catastrophic situation is projected for the study region, with flooding being one of many negative outcomes.

3.2. Satellite monitoring and aerosol transport over glacier region

Fig. 9 shows various transects of CALIPSO across the glacier study region for different seasons. There is a consistent presence of BC, sulfate, and biomass burning, with low values of VDR during winter and fall, which confirms the presence of fine particles. There is support for the presence of dust due to VDR values exceeding 0.5 during spring and summer. These results were further verified and modified by analyzing aerosol subtype observations.

Fig. 9.

Volume depolarization ratio (VDR) values for CALIPSO transects across the glacier region during four seasons.

The aerosol sub-type profiles obtained from CALIPSO during summer, spring, autumn, and winter are shown in Fig. 10. It is evident from the figure that during summer and spring the prominent aerosol types were dust and polluted dust (coarse particles) which reached altitudes of 10 km over the study site. Dust aerosols are present due to long range transport and polluted dust ascribed to anthropogenic events (Yu et al., 2016). CALIPSO observations also showed that polluted dust, polluted continental, clean continental and smoke (fine particles) are the dominant aerosol sub-types during winter, which extend up to 7 km. While during autumn aerosol layer also stretched up to7 kilometers presenting the maximum contribution of smoke (fine particles) and seldem influence of polluted dust (coarse particles) were observed. Prasad et al. (2011) reported the presence of desert dust and other anthropogenic aerosols over the Himalaya region by using CALIPSO observations. Cong et al. (2015) also observed a thick layer of aerosols over the Himalaya and Tibetan plateau reaching over 6 km by using the CALIPSO retrieved aerosol sub-type information for 17 April 2010.

Fig. 10.

CALIPSO retrieved aerosol classification (sub-type profile) over the study region.

The National Oceanic and Atmospheric Administration Hybrid Single Particle Lagrangian Integrated Trajectory (NOAA HYSPLIT) model is useful to determine the origin of air masses impacting a receptor site (Draxler and Rolph, 2003). To determine the mean aerosol transport pathways from the various surrounding regions to the study site, three day air-mass back trajectories were analyzed at 1000 m above the ground by using the HYSPLIT model (Fig. 11). These trajectories are representative of the four seasons. During spring, 62.5% of air masses arrived from the Dasht Desert (Iran), 12.5% from Saratgard, Rajasthan (India), and 25% from central Punjab mainly Cholistan Desert (Pakistan). During summer, 64.3% of air masses arrived from Turkmenistan, 28.6% from Peshawar (Pakistan), while 7.1% are of local origin. During spring and summer there was the maximum contribution of long range transported aerosols confirming the presence of coarse particles over study site. During autumn, 25% of air masses come from Faisalabad (Pakistan), which pass through Amritsar (India) before reaching the receptor site, 12.5% come from Tajikistan and Afghanistan, and 62.5% are of local origin. During winter, 80% of air masses come from Sialkot (Pakistan) and Shakargard (India), 13.3% from Azerbaijan, and 6.7% from Loristan province (Iran). Thus, during autumn and winter the trajectories were mostly local coincident with a stronger influence from fine particles rather than coarse particles. All these results were in good agreement with observation made by CALIPSO. To explore aerosol sources and their possible pathways, similar cluster trajectories were examined in earlier work for other regions (Srivastava et al., 2015; Bibi et al., 2017a,b; Iftikhar et al., 2018).

Fig. 11.

Three day air-mass back-trajectories ending at the receptor site for the four seasons; (a) Spring, (b) Summer, (c) Autumn and (d) Winter, with different colors representing the percentage of air masses.

4. Conclusions

In the present study, remote sensing data are analyzed to examine seasonal and monthly variations in aerosol optical properties between 2004 and 2016 over Gilgit, Skadu, and Diamar (northern Pakistan). The main conclusions of the present work are summarized as follows:

• The variation in aerosol optical properties (AOD, AI, AAOD, SSA) revealed that both scattering and absorbing aerosols are present in the glacier region.

• Frequency distribution shows that a large heterogeneity was observed in the aerosol size distribution with a unimodal distribution in all seasons.

• The increasing trend of AOD, AI and AAOD showed an increase in the aerosol burden over the region. While, the decrease trend in SSA, depicted the increase in absorbing aerosol.

• Contrasting seasonal variations in the VDR suggested that fine particles were dominant during winter and autumn, whereas, coarse particles were dominant during spring and summer.

• The CALIPSO measurement revealed that major aerosol sub-types impacted the study region include dust, polluted dust, clean continental, and smoke.

• HYSPLIT trajectory analysis indicated that air masses reaching the studied sites were from Iran, India, Afghanistan, Turkmenistan, Tajikistan, Azerbaijan, Sialkot, Peshawar, Cholistan Desert, and from local regions.