ENSO Modulation of Interannual Variability of Dust Aerosols over the Northwest Indian Ocean

307
Schematic representation of the conditions associated with ENSO that are important for dust production and transport over northwest Indian Ocean. Enhanced dust production north of the Persian Gulf is attributed to the modulation of precipitation and soil moisture by La Niña, whereas enhanced northwesterly and northeasterly winds during summers following La Niña lead to increased dust production in the Arabian Peninsula and its subsequent transport over the Arabian Sea.

Mineral dust is known to affect many aspects of the climate of the north Indian Ocean (IO). However, what controls its interannual variability over this region is largely unknown. The authors study the mechanism controlling the interannual variability of dust aerosols in the principal dust belts bordering the northwest IO.

It is shown that annual dust activity to the north of the Persian Gulf has an inverse relation with preceding precipitation during October–December and soil moisture during current dust season (April–August). These are in turn remotely controlled by El Niño–Southern Oscillation (ENSO) through the modification of the intensity of convection over the Indo-Pacific warm pool region, which affects moisture flux to the dust sources.

While La Niña leads to a negative precipitation anomaly and more dust generation during the following summer, El Niño is responsible for the opposite. During the summer following La Niña, the air–sea in-teraction leads to a lowering of geopotential height over the Indo-Iranian region, resulting in an increased gradient between the Indo-Iranian region and the surrounding regions. This intensifies the dust-transporting northwesterly and northeasterly winds over the Arabian Peninsula. The dust transport by the intensified low-
level southwesterlies and upper-level westerlies is the main factor responsible for enhanced dust over the open northwest IO during the years following La Niña. The Indian Ocean dipole potentially impacts the variability of dust over the northwest IO by modifying the moisture associated with El Niño.

1. Introduction
The role of airborne mineral dust as a key player in perturbing Earth’s energy budget, the hydrological cycle, and the carbon cycle is being increasingly recognized (IPCC 2013). The northwestern part of the Indian Ocean (IO) is one of the dustiest places on the globe because of its proximity to some of the most arid and semiarid regions of southern and southwestern Asia and the Middle East. Dust emitted from southwestern Asia
and the Middle East (SAME) is often transported eastward across the Arabian Sea (AS) and reaches the Indian subcontinent, which is one of the most densely populated regions of the globe. Dust aerosols over the northwest IO can impact the radiative balance over the
region by leading to a clear-sky (shortwave plus longwave) radiative forcing of 27Wm22 (Zhu et al. 2007). Apossible change in the monsoon precipitation patterns
in response to ‘‘elevated’’ heating within the dust layer has been proposed (e.g., Lau and Kim 2006; Vinoj et al. 2014). A plausible link between mineral dust and phytoplankton
biomass of the AS has also been hypothesized (Patra et al. 2007; Singh et al. 2008; Banerjee and Prasanna Kumar 2014). Every year about 26–531 Tg of dust is estimated to be emitted from the Middle East region (see Huneeus et al. 2011). Considering the high
amount of dust over this region, studies concentrating on the variability of dust have been remarkably few compared to studies of other parts of the globe.
Quite a large number of studies have established the seasonal cycle of dust for the regions bordering the northwest IO, concluding that spring and summer generally
have the highest dust activity (see, e.g., Middleton 1986; Littmann 1991; Li and Ramanathan 2002; Zhu et al. 2007; Yu et al. 2013). Dust activity over the Indian
subcontinent begins during spring, peaks during the month of May, and continues until the southwest monsoon rain establishes itself during July (e.g., Middleton

Dust over the Oman region is associated with the southwest monsoon circulation peaking during July (Ackerman and Cox 1989). Over the Arabian Peninsula, the Tigris–
Euphrates River basin, and the Iran region, dust activity begins during April–May and also peaks in July (Prospero et al. 2002). Overall, around the northwest IO, dust activity attains its maxima during June–July owing to the high heat that leads to the development of
atmospheric instability (Goudie and Middleton 2006). Coupled with this, synoptic-scale weather systems and the complicated nature of the topography of the region
channelize the wind flow to pick up soil dust locally. For example, the Sharav cyclones from the Mediterranean region can often give rise to dusty weather in the northern part of the Arabian Peninsula during winter– spring (Kubilay et al. 2000). The northwesterly shamal
wind is responsible for lifting up dust primarily along the axis of the Persian Gulf (Perrone 1979; Rao et al. 2003; Vishkaee et al. 2012). The Levar in the Sistan region of
Iran often results in dusty weather continuously for months (Kaskaoutis et al. 2014). It has been seen that the intensification of the pressure gradient between the Mediterranean region and the Indo-Iranian region leads to enhanced dust activity in Iraq and the Arabian Peninsula
(Rao et al. 2003; Awad and Mashat 2013). Thus, at the seasonal scale, dust activity and the controlling  factors have been studied reasonably well. The understanding of what controls the interannual variability of dust, however, eludes us in spite of some isolated
efforts. Rahul et al. (2008) have shown that in general there is higher aerosol loading over the AS during the years when the Indian summer monsoon is normal compared to those years when there is drought over India. They attributed this to the easterly wind
anomaly during the drought years that prevented transport of sea salt, thereby leading to less rainfall over India. El Niño conditions have been associated with increased dust transport from the Middle East to the Indian subcontinent by the strengthened westerlies
(Abish and Mohanakumar 2013). It is important to note that the effect of El Niño on the rainfall anomaly (an important factor for dust production) of the Indian subcontinent and SAME are contrasting. While El Niño years are usually associated with drought over India
(e.g., Ropelewski and Halpert 1987; Kumar et al. 2006), they result in wet conditions over SAME (Nazemosadat and Ghasemi 2004; Mariotti 2007). Gautam et al. (2009)
have shown that unusually low dust over northwestern India in 2007 is related to high precipitation in the Thar Desert during the previous winter. Recently, Yu et al.
(2015) concluded that La Niña conditions favor springtime dust activity over the Arabian Peninsula and related the decadal variability of dust activity to rainfall over North Africa and sea surface temperature over the Mediterranean Sea. Over the Sistan region of Iran, no
significant relation between summertime El Niño– Southern Oscillation (ENSO) and dust storms has been discerned (Kaskaoutis et al. 2014). Thus, there is no clear consensus or understanding of the global teleconnection patterns that might control dust activity over
the northwest IO. Development of such an understanding is crucial to explain the interannual variability of dust load over a region and how the global climate change may affect the same.
This is the motivation of the present study wherein we seek to address what large-scale circulation systems related  to the climate modes orchestrate the interannual
variability of dust activity in the regions surrounding the northwest IO and its subsequent transport. The two most important climate modes that have the potential to impact the north IO are ENSO (Trenberth 1997; Kumar et al. 2006) and the Indian Ocean dipole (IOD; Saji et al. 1999). While ENSO has its birth in the tropical Pacific Ocean, IOD is an equivalent of ENSO in the tropical IO. Most often, it has been seen that ENSO and IOD work hand in hand, each complementing the effect of the other (Meyers et al. 2007). For example, during the positive phase of ENSO and IOD, there is an anomalous cooling of the sea surface temperature (SST) over the eastern tropical IO and suppression of convective activity over this region and vice versa. Also, the tropical west IO experiences anomalous warming (cooling) during positive (negative) IOD (Saji et al. 1999). In this study we try to unravel whether ENSO and IOD have the potential to significantly affect dust activity around
the northwest IO.

2. Data and methods
The time series of dust aerosols in the atmosphere over a region is perhaps best represented by ultraviolet absorbing aerosol index (AI), which, although discontinuous
at times, has been giving a semiquantitative indication of dust load over both land and ocean since the late 1970s. AI is an indicator of how much absorbing and nonabsorbing spectral radiance ratios in an atmosphere with aerosols differs from that of a purely molecular
atmosphere (Herman et al. 1997). Daily AI data were obtained from the Total Ozone Mapping Spectrometer (TOMS), version 8, on board the Nimbus-7 satellite for the period from January 1979 to December 1990 and onboard the Earth Probe (EP) satellite for the
period from January 1997 to December 2001 at a spatial resolution of 1.008 3 1.258. Data from the Ozone Monitoring Instrument (OMI), version 3, onboard the Aura 1288 JOURNAL OF CLIMATE VOLUME 29 satellite were used for a more recent period encompassing
January 2005–December 2013 at a spatial resolution of 1.008 3 1.008. The data for the period 1991–92 were not considered in order to avoid contamination of
the dust signal by volcanic aerosols due to the eruption of Mount Pinatubo. To ascertain whether a similar contamination occurred as a result of the El Chichóneruption in 1982, we have carried out the analysis with and without data from 1982 and 1983. The results did not
show any marked change. Hence, we have retained the data for the same period. Also, the data for the period 2002–04 were discarded because of a calibration problem with TOMS–EP associated with a wavelength dependent shift in the optical properties of the front
scan mirror. This calibration problem has resulted in increased noise level in the AI data, which affected the nature of the annual cycle of AI as well as its amplitude (Kiss et al. 2007). The daily AI data for a total of 26 years were screened to enumerate the percentage of days eachyear when AI exceeded a threshold value of 1.5 (AI-1.5). This gave an indication of the percentage of days each year when there was a significant amount of dust activity
in the region. Taking other thresholds for AI such as 1.0 or 2.0 has yielded more or less the same results with some increase and decrease, respectively, in the area
extent of the main dust activity regions. There are two major caveats when using AI data:
1) although dust is an ultraviolet-absorbing aerosol, black carbon can also absorb radiation in the ultraviolet spectrum, and 2) the AI retrieval is highly dependent on
the height of the dust aerosols, with dust in the upper layer being more effectively detected than that near the surface (Herman and Celarier 1997). There is a possibility
that AI may actually represent dust that is being transported. However, several studies have shown thatover the dust source regions the frequency of days when AI attains high values is much greater than in other regions (e.g., Prospero et al. 2002; Ginoux et al. 2012). We
justify the use of AI-1.5 because the study is concerned with known arid and semiarid dust source regions and because it offers a comparatively long time period over which AI data are available, which is essential for climate studies. ENSO was represented using the Extended Reconstructed SST (ERSST) Niño-3.4 index (http://www.cpc.ncep.noaa.gov/data/indices/ersst3b.nino.mth.81-10.ascii) for October December (OND) of each year for the period 1978–2013. The Niño-3.4 index uses SST in the region 58N–58S, 1708–1208W. A year was designated as an El Niño (warm) year when the Niño-3.4 index exceeded a value of 1.0. Years having Niño-3.4 indices less than 21.0 were taken as La Niña (cold) years. Those years when the Niño-3.4 index was between21.0 and 1.0 were regarded as normal years. The dipole mode index (DMI), representing the phase and amplitude of IOD, averaged over September–November (SON), was determined
using ERSST (Xue et al. 2003; Smith et al. 2008) data by calculating the difference in SST anomaly between western tropical IO (108S–108N, 508–708E) and eastern tropical IO (58–108S, 908–1108E). The region chosen for this calculation is very similar to the boxes
used by Saji et al. (1999) and Meyers et al. (2007) for defining IOD. A year was regarded as a positive IOD when the DMI was above 0.5. The lowest DMI of20.5 is seen during 1996. A normal year was regarded as one that had a DMI between 20.5 and 0.5. Considering the
lesser intensity of negative IOD, we have mostly concentrated on the positive phase of IOD.
Climate Prediction Center modeled monthly soil moisture data were used at spatial resolution of 0.58 30.58 for the period 1978–2013. Soil moisture is calculated by means of a one-layer water balance model using observed monthly precipitation over land and temperature
(Fan and van den Dool 2004). Monthly precipitation data over land were obtained from the Global Precipitation Climatology Centre at a spatial resolution of 0.58 3 0.58 for the period 1978–2010 (Schneider et al. 2014). Additionally, the climate data record (CDR)
normalized difference vegetation index (NDVI), obtained from the National Oceanic and Atmospheric Administration’s (NOAA) Advanced Very High Resolution Radiometer (AVHRR), was also used for the period 1982–2013 (Pedelty et al. 2007). NDVI refers to the ratio between red and infrared reflectance and is an indicator of the vegetation health. The data available as a daily product at 0.058 3 0.058 spatial resolution have been regridded to 0.258 3 0.258 for the ease of computation. Data for wind vectors, specific humidity,
geopotential height, and vertical velocity were obtained from the National Centers for Environmental Prediction–National Center for Atmospheric Research
(NCEP–NCAR) reanalysis (Kalnay et al. 1996), and outgoing longwave radiation (OLR) data were obtained from NOAA (Liebmann and Smith 1996). Partial correlations  between the datasets were calculated to determine the signal due to ‘‘pure’’ ENSO and IOD using
the method similar to Yamagata et al. (2004), and the significance of the correlation coefficients were examined using a two-tailed t test.

3. Results and discussion
Our approach is to first give an overview of the spatial  pattern of AI-1.5 distribution and the nature of temporal evolution of AI within SAME. We then explore how the interannual variability of AI-1.5 might be related to ENSO and IOD.

FIG. 1. Shadings show the climatological (1979–2013) distribution of (a) AI-1.5, (b) precipitation, and (c) soil moisture. The black contours are their standard deviations. The vectors and the white dashed line in (a) show the wind pattern at 850 mb and the mean position of the ITCZ determined as the location where the northward component of the wind becomes zero during April–August, respectively. The black box in (a) represents the SAME region that is used for further study. (d) The temporal evolution of AI, precipitation, and soil moisture over SAME. The black curve shows the daily climatology of AI, with standard deviations of AI represented by the gray shaded region, and the horizontal dashed line indicates where the value of AI is 1.5. The small squares indicate the monthly climatologies of precipitation plotted along with their error bars. The large black rectangles are the monthly mean climatologies of the soil moisture. The numbers below the rectangles indicate the standard deviations of soil moisture.
FIG. 1. Shadings show the climatological (1979–2013) distribution of (a) AI-1.5, (b) precipitation, and (c) soil moisture. The black contours are their standard deviations. The vectors and the white dashed line in (a) show the wind pattern at 850 mb and the mean position of the ITCZ determined as the location where the northward component of the wind becomes zero during April–August, respectively. The black box in (a) represents the SAME region that is used for further study. (d) The temporal evolution of AI, precipitation, and soil moisture over SAME. The black curve shows the daily climatology of AI, with standard deviations of AI represented by the gray shaded region, and the horizontal dashed line indicates where the value of AI is 1.5. The small squares indicate the monthly climatologies of precipitation plotted along with their error bars. The large black rectangles are the monthly mean climatologies of the soil moisture. The numbers below the rectangles indicate the standard deviations of soil moisture.

Finally, we try to understand the physical mechanism through which ENSO and IOD might affect the interannual variability of AI-1.5 over the northwest IO.

For more details please Click

– PRIYANKA BANERJEE AND S. PRASANNA KUMAR
National Institute of Oceanography, Council of Scientific and Industrial Research, Dona Paula, Goa, India

This is already published in Journal of Climate, DOI: 10.1175/JCLI-D-15-0039.1 

NO COMMENTS

LEAVE A REPLY