The sampling site, Shillong (25.67˚N, 91.91˚E, 1064 m ASL), located in the North-Eastern part of India, is characterized by a high annual rainfall (~2200 mm). While the high precipitation help retain a clean atmosphere via wet scavenging, occasional long range transport of pollutants many a time shows up clearly over the background in the atmospheric trace gas and aerosol measurements [6] [10] , making it an ideal site for studying long range pollution transport episodes. Its latitudinal location and high elevation provide it with a subtropical climate with mild summers and chilly to cold winters. The monsoon arrives by June and rains almost until the end of August, with low intensity precipitations sometimes showing up on either side of these months. The window of opportunity for most of the aerosol and trace gas measurements at this region is Jan - Mar every year, when hardly any precipitation occurs.

For the ambient SO₂ measurements, a primary UV fluorescence SO₂ monitor (Thermo-43i TLE) with a lower detection limit of 0.05 ppbv was used (see http://www.thermoscientific.com/content/tfs/en/product/enhanced-trace-level-so-sub-2-sub-analyzer-model-43-i-i-i-tle.html ). Also see, [11] [12] [13] for detailed evaluations of similar systems. The dynamic gas calibrator (Thermo-146i) fed with a standard SO₂ gas (2 ppmv with N₂ balance gas, Spectra, USA) facilitated the routine onsite calibrations. See [6] for an elaborate discussion on the instrumentation aspects of the SO₂ monitoring.

The experimental setup for the SO 4 2 − measurements comprised periodic collection of fine mode aerosol samples using a high volume air sampler (Thermo) and the subsequent measurement of inorganic anions via ion chromatography. The aerosol samples (PM_2.5) were collected on Whatman cellulose filters (200 × 250 mm²) loaded in the air sampler (flow rate, 1.12 m³/min). In general, the procedure involved collecting two aerosol samples (sampling duration ~6 hrs) during daytime and one sample (sampling duration ~12 hrs) during night time. A quarter section of each filter was then soaked in 50 mL Milli-Q water (18.2 MΩ resistivity) for 4 hrs with ultrasonication performed for 20 minutes (in steps of 5 minutes). This procedure essentially extracts all the water soluble ionic species (WSIS) from the filter. The water extract is then passed though an ion chromatograph (Dionex, Model 2000i/SP with CDM-3 conductivity detector, see [14] [15] for a detailed discussion of the system) to measure the inorganic anions (Cl⁻, NO 3 − and SO 4 2 − ). The system employs Ionpac AS14A analytical column and AG14A guard column, in conjunction with an anion self-regenerating suppressor (ASRS) in recycle mode, using 8.0 mM Na₂CO₃ - 1.0 mM NaHCO₃ as eluent to separate the anions. The measured values were then further corrected for procedural blanks (comprising blank filters and analytical reagents).

The GEOS-Chem global 3-dimensional chemical transport model (v8-03-01; http://acmg.seas.harvard.edu/geos/ ) with HO_x-NO_x-VOC-ozone chemistry [16] [17] have been employed for this study. For the present simulations, the model employed GEOS-5 assimilated meteorology having a temporal resolution of 6-h (3-hour resolution for surface fields and mixing depths) and a horizontal resolution of 0.5˚ latitude × 0.667˚ longitude, with 72 vertical levels, regridded to 4˚ × 5˚. The elaborate model evaluations can be found in [17] [18] [19] [20] . See, [21] for the aerosol sources and processes, [19] [22] for modifications related to dust and sea salt, [23] for the wet deposition of soluble aerosols and gases, and [24] for dry deposition via the standard resistance-in-series scheme.

The model spin up time was kept sufficiently long (12 months, from Jan 2008 to Jan 2009) to remove any influence of initial conditions. The simulations were then performed for Jan 2009 - Apr 2010 at 4˚ × 5˚ resolutions with the standard input.geos file (unless mentioned otherwise) distributed with the GEOS-Chem codes (v8-03-01), that define the input parameters and emission inventories-EMEP [25] , BRAVO [26] , EDGAR [27] , Streets inventory [28] , CAC ( http://wiki.seas.harvard.edu/geos-chem/index.php/CAC_anthropogenic_emissions ), and EPA/NEI05. The ND49 diagnostics were turned ON in the model to generate the time series data of the various species. The primary analysis of the model outputs were performed with the global atmospheric model analysis package (GAMAP) (Version 2.15; http://acmg.seas.harvard.edu/gamap /).

When the transient long range transport brought SO₂ plumes to the sampling site, the S-ratios, derived from time series SO 4 2 − and SO₂ measurements, were seen to be unusually low (median value, 0.03). The anomalous SO₂ oxidation efficiency patterns can be very well seen in the time series S-ratios for the different sampling intervals (Figure 3), which showed a dip during 11^th and 12^th March

2009 followed by a steady ascend to reach its maximum through to 20^th followed by a decrease. The lowest ratio recorded was 0.008 (at 0159 hrs, 12^th Mar) and the highest was 0.11 (at 1620 hrs, 20^th Mar). Throughout this sampling, the S-ratios remained anomalously low.

Similarly, the measured time series SO₂ and SO 4 2 − (Figure 4) exhibited many unusual features. During 9^th, 10^th and 11^th March, the SO₂ variability was governed predominantly by long range transport influx with minimal planetary boundary layer (PBL) height diurnality influence. An interesting observation for these days is that the SO₂ and SO 4 2 − peaks were almost in anti-phase with each other, which may be attributed to the differing mobility pattern and loss mechanisms for SO₂ (gas) and SO 4 2 − (particulate matter), to result in apparent transit time/concentration mismatches in reaching the sampling site.

Observations similar to the one here―viz., ‘the poor SO 2 -SO 4 2 − correlations/phase-agreements in long range transported air masses’―were reported by [9] when plumes from an accidental oil fire event [31] were detected at their high altitude sampling site Mt. Abu, India, and had then projected the scope for further systematic simulation/field studies on the topic.

From 12^th March onwards, the SO₂ variability is seen to have a more PBL height diurnality influence. During 12^th to 20^th also, the SO₂ and SO 4 2 − peaks are in anti-phase, and may be explained by a non-availability of sufficient

enough OH radicals, which were possibly heavily utilized for the oxidation of large amounts of SO₂ [6] and other pollutants [10] in the initial days, to lead to a ‘suppressed oxidation condition’ for the local atmosphere―as evidenced by the anomalously low S-ratios.

During 21^st to 26^th, when SO₂ reduced significantly compared to the initial sampling days, the SO₂ and SO 4 2 − time series variations came in-phase with each other indicating the initiation of a regain of the oxidizing power of the atmosphere via a buildup of sufficient OH.

Figure 5 depicts the S-ratio variabilities as a function of SO₂ during the different sampling intervals in March 2009. The ratio is seen to be high at low SO₂ concentrations and decreased monotonically with increasing SO₂, asymptotically approaching a value of 0.01 at high SO₂ concentrations―an observation quite similar to the one reported by [32] . But they reported the S-ratio to asymptotically approach a value of 0.1 at high SO₂ concentrations and suggested such a behavior to be consistent with a conversion of about 10% of the fuel sulphur to SO 4 2 − in the source region followed by (1) oxidation of the SO₂ to SO 4 2 − as the plume transported downwind (2) dilution of the plume with air containing a high fraction of the sulphur present as SO 4 2 − or (3) a combination of these two processes.

To further explore the notion of ‘suppressed oxidation condition’ over the sampling region and assess the factors contributed to the anomalous low S-

ratios, simulation studies with the chemical transport model also were performed, and is detailed below.

The GEOS-Chem (v8-03-01) runs were performed with the inventories discussed previously―employing the standard input.geos file―to generate time series SO₂ and SO 4 2 − for the 4˚ × 5˚ grid cell containing the sampling site, and the S-ratios were derived for March 2009. Similar to the experimental observations, the simulations also showed anomalous low S-ratios during 10^th to 16^th March, and the time series SO₂ and SO 4 2 − variabilities exhibited poor phase agreement (Figure 6).

To study the scenario further, time series OH radicals were generated (Figure 7), from the model by turning ON the ND49 diagnostics. The OH decreased steadily from 7^th March 2009 to reach the lowest of the month on 16^th, followed by an ascend. Further, the time series OH normalized to time series SO₂, referred here as OH_specific gave ‘consistently suppressed values’ during 10^th to 16^th (Figure 8), thus explaining the observed ‘S-ratio anomaly’ for the period.

Having seen ‘suppressed oxidation conditions’ in both experiments and simulations, it is worth assessing the contributions from factors such as ‘Transport’, ‘dry deposition’ and ‘dust load’ to the scenario. Sensitivity simulations―similar to the ones performed by [9] ―can help resolve some of these factors and are detailed in the following sections 3.2.1 to 3.2.3. From these, the

effect of switching OFF different ‘processes’ on the SO₂, SO 4 2 − , and S-ratios are assessed via the equation:

Δ Species ( process ) = ( Species ' − Species ) / Species × 100 % (2)

where:

ΔSpecies_(process) = Percent Difference in Species concentration/value in the absence of the ‘Process’

Species = Species concentration/value from the baseline run employing the standard input.geos file

Species' = Species concentration/value from the sensitivity run with the particular ‘process’ turned OFF

The ‘Species’ mentioned here are SO₂, SO 4 2 − , and S-ratio respectively.

The degree of temporal and spatial variability of tropospheric OH and its effects on the species life time has been a topic of debate [52] - [57] . In the wake of the very limited direct OH radical measurements available for the region, GEOS-Chem model simulations can help provide significant insights into the geographical distribution pattern of OH.

Figure 13 shows the mean OH concentrations for different months of the year 2009, from the GEOS-Chem model. A clear OH minimum is seen over the region surrounding (20˚N, 95˚E) spanning parts of the north-east India and the adjacent regions to the south-east of it. This minimum is prevalent throughout

the year, though the magnitude and the area of influence have a seasonality to it. This model prediction of an OH minimum can have significant implications to the present understanding on the oxidizing power of the regional atmosphere and hence on the air quality, especially during the long range pollution transport episodes associated with spring time cold air outbreaks. For example, [58] from their model studies showed that an ‘OH minimum’ in west Pacific could reduce the upper free tropospheric aerosol surface area density by up to 25% (due to less efficient conversion of SO₂ into sulphate), while that in the lowermost stratosphere increase by more than 5% (due to increased SO₂ transport into the stratosphere and conversion into sulphate at higher altitudes).

It is clear that the OH minimum conditions reflected in the model simulations, for the region, could well be the possible explanation for the suppressed S-ratio values during the plume transport episode in March 2009. The ‘OH minimum’ in the model simulations along with the ‘anomalous low S-ratios’ in the field measurements, underlines the impending need for systematic field measurements of OH in north-east India and the adjoining regions―in the wake of the region’s growing anthropogenic emissions.

The simultaneous measurements of ambient SO₂ and SO 4 2 − were again performed at our sampling site, in January 2010. This sampling period didn’t see any major long range transports [6] and the air mass back trajectories extended mostly to the Indo-Gangetic plane (figures not shown). For this month, no any kind of anomalies were seen in the SO₂ oxidation efficiencies, and the S-ratios (Figure 14) were well within the acceptable limits, with a monthly median value of 0.32. The S-ratio values for this month are comparable to those reported by [9] , for their western Indian site, Mt. Abu. A very good phase-agreement is seen between the SO₂ and SO 4 2 − time series during the different sampling intervals for this month.

This observation of an in-phase varying time series SO 2 -SO 4 2 − and a monthly median S-ratio value on par with those reported from western India [9] , would possibly imply that under the present local emission conditions of the region, the oxidizing power of the regional atmosphere is still somewhat sufficient to have normal oxidation rates for the trace gases including SO₂, which becomes insufficient during major long range pollution transport episodes―peaking in the spring time [6] [10] when cold air outbreaks cause significant pollutant transport from far flung regions.