Variations in O 3 , CO , and CH 4 over the Bay of Bengal during the summer monsoon season : shipborne measurements and model simulations

We present shipborne measurements of surface ozone (O3), carbon monoxide (CO), and methane (CH4) over the Bay of Bengal (BoB), the first time such measurements have been performed during the summer monsoon season, as a part of the Continental Tropical Convergence Zone (CTCZ) experiment during 2009. O3, CO, and CH4 mixing ratios exhibited significant spatial and temporal variability in the ranges of 8–54 nmol mol−1, 50–200 nmol mol−1, and 1.57– 2.15 μmol mol−1, with means of 29.7± 6.8 nmol mol−1, 96± 25 nmol mol−1, and 1.83± 0.14 μmol mol−1, respectively. The average mixing ratios of trace gases over BoB in air masses from central/northern India (O3: 30± 7 nmol mol−1; CO: 95± 25 nmol mol−1; CH4: 1.86± 0.12 μmol mol−1) were not statistically different from those in air masses from southern India (O3: 27± 5 nmol mol−1; CO: 101± 27 nmol mol−1; CH4: 1.72± 0.14 μmol mol−1). Spatial variability is observed to be most significant for CH4 with higher mixing ratios in the air masses from central/northern India, where higher CH4 levels are seen in the SCIAMACHY (SCanning Imaging Absorption spectroMeter for Atmospheric CartograpHY) data. O3 mixing ratios over the BoB showed large reductions (by ∼ 20 nmol mol−1) during four rainfall events. Temporal changes in the meteorological parameters, in conjunction with O3 vertical profile, indicate that these low-O3 events are associated with downdrafts of free-tropospheric O3-poor air masses. While the observed variations of O3 and CO are successfully reproduced using the Weather Research and Forecasting model with Chemistry (WRF-Chem), this model overestimates mean concentrations by about 6 and 16 % for O3 and CO, respectively, generally overestimating O3 mixing ratios during the rainfall events. An analysis of modelled O3 along air mass trajectories show mean en route O3 production rate of about 4.6 nmol mol−1 day−1 in the outflow towards the BoB. Analysis of the various tendencies from model simulations during an event on 10 August 2009, reproduced by the model, shows horizontal advection rapidly transporting O3-rich air masses from near the coast across the BoB. This study fills a gap in the availability of trace gas measurements over the BoB and, when combined with data from previous campaigns, reveals large seasonal amplitude (∼ 39 and ∼ 207 nmol mol−1 for O3 and CO, respectively) over the northern BoB.

ozone (O 3 ), carbon monoxide (CO), and methane (CH 4 ) over the Bay of Bengal (BoB), the first time such measurements have been performed during the summer monsoon season, as a part of the Continental Tropical Convergence Zone (CTCZ) experiment during 2009.O 3 , CO, and CH 4 mixing ratios exhibited significant spatial and temporal variability in the ranges of 8-54 nmol mol −1 , 50-200 nmol mol −1 , and 1.57-2.15µmol mol −1 , with means of 29.7 ± 6.8 nmol mol −1 , 96 ± 25 nmol mol −1 , and 1.83 ± 0.14 µmol mol −1 , respectively.The average mixing ratios of trace gases over BoB in air masses from central/northern India (O 3 : 30 ± 7 nmol mol −1 ; CO: 95 ± 25 nmol mol −1 ; CH 4 : 1.86 ± 0.12 µmol mol −1 ) were not statistically different from those in air masses from southern India (O 3 : 27 ± 5 nmol mol −1 ; CO: 101 ± 27 nmol mol −1 ; CH 4 : 1.72 ± 0.14 µmol mol −1 ).Spatial variability is observed to be most significant for CH 4 with higher mixing ratios in the air masses from central/northern India, where higher CH 4 levels are seen in the SCIAMACHY (SCanning Imaging Absorption spectroMeter for Atmospheric CartograpHY) data.O 3 mixing ratios over the BoB showed large reductions (by ∼ 20 nmol mol −1 ) during four rainfall events.Temporal changes in the meteorological parameters, in conjunction with O 3 vertical profile, indicate that these low-O 3 events are associated with downdrafts of free-tropospheric O 3 -poor air masses.While the observed variations of O 3 and CO are successfully reproduced using the Weather Research and Forecasting model with Chemistry (WRF-Chem), this model overestimates mean concentrations by about 6 and 16 % for O 3 and CO, respectively, generally overestimating O 3 mixing ratios during the rainfall events.An analysis of modelled O 3 along air mass trajectories show mean en route O 3 production rate of about 4.6 nmol mol −1 day −1 in the outflow towards the BoB.Analysis of the various tendencies from model simulations during an event on 10 August 2009, reproduced by the model, shows horizontal advection rapidly transporting O 3 -rich air masses from near the coast across the BoB.This study fills a gap in the availability of trace gas measurements over the BoB and, when combined with data from previous campaigns, reveals large seasonal amplitude (∼ 39 and ∼ 207 nmol mol −1 for O 3 and CO, respectively) over the northern BoB.

Introduction
Tropospheric ozone (O 3 ) is the third most important greenhouse gas, contributing to global warming and climate change with a radiative forcing of 0.40 ± 0.20 Wm −2 (IPCC, 2013).O 3 is also a pivotal trace gas in tropospheric chemistry, as it is a major source of hydroxyl radical (OH), which removes most of the organic compounds and pollutants from the atmosphere and controls the oxidation capacity of the troposphere (e.g.Brasseur et al., 1999;Finlayson-Pitts and Pitts, 2003;Seinfeld and Pandis, 2006).Further, enhanced concentrations of surface O 3 have detrimental effects on human health and vegetation (Heagle, 1989;Seinfeld and Pandis, 2006).Approximately 80 % of tropospheric O 3 is produced by in situ photochemical reactions in the presence of nitrogen oxides (NO x = NO + NO 2 ) involving the precursor gases of methane, non-methane hydrocarbons (NMHCs), and CO (Fishman et al., 1979;Crutzen et al., 1999;Seinfeld and Pandis, 2006).The remaining 20 % of tropospheric O 3 is attributed to intrusions of stratospheric air during frontal activities or to tropopause folding events (Lelieveld and Dentener, 2000;Sprenger et al., 2007).Depending upon meteorological conditions and the availability of the aforementioned precursors, a net production or destruction of O 3 prevails.The average lifetime of O 3 is about 1 week in the lower troposphere, which leads to large variability in its spatial and temporal distributions, as compared to the long-lived greenhouse gases.The budget of tropospheric O 3 and its implications for human health, crop yields, and climate are, however, not yet well quantified, especially over regions in Asia.This is mainly due to a lack of in situ measurements (e.g.Cooper et al., 2014;Monks et al., 2015).
Carbon monoxide (CO) is an indirect greenhouse gas which also has adverse effects on the health of humans and animals (WHO, 1999).Although it does not have a direct greenhouse effect like methane and carbon dioxide, its role in atmospheric chemistry is estimated to cause an indirect radiative forcing of 0.23 (0.18-0.29)Wm −2 (IPCC, 2013).The major sources of CO are fossil fuel combustion, biomass burning, and oxidation of hydrocarbons such as CH 4 and isoprene (e.g.Jacob, 1999;Bergamaschi et al., 2000;Seinfeld and Pandis, 2006).
Methane (CH 4 ) is one of the major greenhouse gases, with a direct radiative forcing of 0.48 ± 0.05 Wm −2 (IPCC, 2013).This gas plays a major role in the climate and in atmospheric chemistry.CH 4 is emitted from variety of natural and anthropogenic sources (Jacob, 1999) and is removed primarily through its reaction with OH radicals (Fung et al., 1991;Seinfeld and Pandis, 2006).
The marine regions adjoining South Asia have been observed to have elevated levels of surface O 3 due to the outflow of continental pollution (Lawrence and Lelieveld, 2010) and minimal chemical loss by titration (e.g.Lal and Lawrence, 2001;Ojha et al., 2012).Suggested sources for this elevated O 3 and other trace gases observed over the marine regions surrounding India are anthropogenic, biomass burning, and biogenic emissions over continental India (Naja et al., 2004;Lawrence and Lelieveld, 2010;Nair et al., 2011;David et al., 2011).The air masses influenced by continental emissions undergo chemical transformation, including O 3 production, during their transport to the cleaner marine regions.In situ measurements over the marine regions are required to understand the effects of direct outflow, en route chemical transformation, and the chemistry in the transported air masses (Lawrence and Lelieveld, 2010, and references therein).
The experiments that have been conducted to date over the marine environment adjacent to the Indian region have revealed considerable spatial heterogeneity in the distribution of trace gases and aerosols, influences from source regions such as the Indo-Gangetic Plain (IGP), and radiative impacts (Nair et al., 2010(Nair et al., , 2011;;David et al., 2011;Mallik et al., 2013;Moorthy et al., 2009).Observations made during the Indian Ocean Experiment (INDOEX; Lal and Lawrence, 2001) and model simulations (Ojha et al., 2012) both found the O 3 mixing ratios over these remote marine regions to be even higher than those over the upwind continental regions due to complex O 3 chemistry.Lawrence and Lelieveld (2010) provided a detailed review of the outflow of trace gases and aerosols from South Asia to the surrounding marine regions.Transport of air masses between the Indian subcontinent and the adjacent marine regions has strong seasonal dependence associated with the monsoonal circulation (e.g.Kumar et al., 2015a).
The marine environment of the Bay of Bengal (BoB), the largest bay in the world, is surrounded by landmasses on three sides, making it highly suitable to observe enhanced concentrations of trace species.Further, seasonal changes in synoptic winds make this a unique region to study variations in trace species due to transport and en route photochemistry.Considering the aforementioned special characteristics of the BoB, as well as the considerable heterogeneity of trace gas and aerosol distribution, in situ measurements covering large areas are essential for investigating the distribution of pollutants and the controlling processes.Extensive in situ measurements of various trace gases over the BoB have been conducted in the following field campaigns: IN-DOEX during the winter months of 1998 and 1999 (Lelieveld et al., 2001;Mühle et al., 2002); the Integrated Campaign for Aerosols, gases, and Radiation Budget (ICARB) during the March-May (pre-monsoon season) of 2006 (Nair et al., 2011;Srivastava et al., 2011Srivastava et al., , 2012)); the winter-ICARB (W-ICARB) during December-January 2009 (Girach andNair, 2010, 2014;David et al., 2011); the Bay of Bengal Experiment (BOBEX)-I during February-March 2001 (Lal et al., 2006); the Bay of Bengal Process Studies (BOBPS) during September-October 2002 (Sahu et al., 2006); BOBEX-II during February 2003 (Lal et al., 2007); and the other campaign conducted during October-November 2010 (Mallik et al., 2013).
Although earlier studies have covered the spatio-temporal distribution of trace gases during most seasons over the BoB, there is still a lack of observations over the BoB during the summer monsoon season (June-August).The Asian summer monsoon circulation provides a pathway for pollution transport into the stratosphere (Randel et al., 2010), and observations taken during monsoon season capture a time of high water-vapour loading over the BoB.Deep convection during the summer monsoon can uplift boundary layer pollution to higher altitudes; the pollution is then distributed over a larger region, thereby influencing air quality and climate over much larger regions (Lawrence and Lelieveld, 2010), extending as far as, for example, over the Mediterranean (e.g.Lelieveld et al., 2002;Scheeren et al., 2003).Such in situ measurements are also essential given the fact that satellite remote sensing of boundary layer O 3 has relatively higher uncertainty.The uncertainties in satellite retrievals of trace species are particularly high during the summer monsoon season, as the view of satellite instruments is frequently obscured by thick clouds.
In the present paper, the ship-based measurements of surface O 3 , CO, and CH 4 over the BoB are presented for the summer monsoon season of 2009.These observations were carried out as a part of the Continental Tropical Convergence Zone (CTCZ) experiment (http://www.incois.gov.in/portal/datainfo/pdctcz.jsp)under the Indian Climate Research Programme (ICRP) of the Government of India.In this study, the spatial and temporal variations of O 3 over the BoB and the effects of transport are analysed.These observations are compared with simulations from a regional model, Weather Research and Forecasting coupled with Chemistry (WRF-Chem).The sharp reductions observed in O 3 during rainfall events are investigated in greater detail.

The cruise track and background conditions
Figure 1 shows the cruise track of the Oceanic Research Vessel (ORV) Sagar Kanya during the CTCZ campaign (cruise number SK 261).The arrows marked on the track show the direction of the ship, which sailed from Chennai (80.The average wind pattern at 925 hPa (NCEP/NCAR reanalysis; http://www.esrl.noaa.gov/psd)during the cruise period is shown in Fig. 1.The prevailing westerly and southwesterly winds transport O 3 and its precursors from the Indian landmass to the BoB during the study period.The spatial distribution of emissions of NO x , an O 3 precursor gas, is also shown as colour map in Fig. 1.NO x emissions are obtained from the Intercontinental Chemical Transport Experiment Phase B (INTEX-B) inventory (Zhang et al., 2009), which is representative of the year 2006.NO x emissions are relatively higher over parts of eastern and southern India as compared to central India.The square tagged as Thiruvananthapuram shows the location corresponding to the measurements shown in Fig. 9.

Experimental details and data
Surface O 3 measurements were carried out using an online ultraviolet (UV) photometric ozone analyzer (model O3 42), manufactured by Environnement S.A, France.The analyser utilises the absorption of UV radiation by O 3 molecules at 253.7 nm and derives O 3 mixing ratios using the Beer-Lambert law.This UV absorption-based analyser has an uncertainty of about 5 % (Tanimoto et al., 2007), corresponding to ∼ 1.5 nmol mol −1 for the observed range of O 3 .Zero noise of the instrument is 0.5 nmol mol −1 .The instrument has a lower detection limit of 1 nmol mol −1 and a linearity of ±1 %.An individual measurement is performed at a minimum response time of 10 s.The analyzer was operated on auto-response mode, whereby responses could be 10-90 s depending upon changes in O 3 mixing ratios.However, data were recorded continuously at 5 min intervals.
CO measurements were made using an online CO analyzer (model CO12 Module) manufactured by Environnement S.A, France.This instrument works on the principle of nondispersive infrared (NDIR) absorption by CO molecules at the wavelength of 4.67 µm.The instrument has a lower detection limit of 50 nmol mol −1 , a linearity of 1 %, and a response time of 40 s.The overall uncertainty in hourly CO measurements is estimated to be ∼ 10 % at a CO value of 150 nmol mol −1 (Sawa et al., 2007;Tanimoto et al., 2007).
Air was drawn from a height of approximately 15 m above the sea surface through a Teflon tube.Before and after the cruise, both analyzers were calibrated, with calibration factors not found to be significantly changed.The calibrations of both analysers were carried out using appropriate calibration standards traceable to NIST and a multi-channel calibrator, following the procedure mentioned in the manuals of analysers.While the O 3 analyser was calibrated for mixing ratios of 30 nmol mol −1 , the CO analyser was calibrated for mixing ratios of 1.1 µmol mol −1 .Meteorological parameters such as pressure, temperature, and relative humidity were measured continuously on board the ship.Trace gas measurements affected by the ship exhaust were identified and discarded using on-board wind direction and NO x measurements.
In addition, a total of 29 air samples were collected in 1 L glass flasks during the cruise and were analysed for CH 4 using a gas chromatograph (GC) coupled with a flame ionisation detector (FID), as described in Tiwari and Ravi Kumar (2011).These CH 4 measurements are traceable to the WMO standard scale.CH 4 standards were obtained from the WMO Central Calibration Laboratory (CCL) at the National Oceanic and Atmospheric Administration (NOAA)/Earth System Research Laboratory (ESRL)/Global Monitoring Division (GMD), located in Boulder, Colorado, USA.The precision for CH 4 measurements was approximately ±0.1 µmol mol −1 .A detailed description of the analytical procedure for CH 4 measurement and calibration of GC is given in Ravikumar et al. (2014).
To further study the observed low-O 3 events over BoB, measurements made at Thumba, Thiruvananthapuram, are used as a case study.Using the same O 3 analyzer as the one used for surface O 3 measurements over BoB, continuous measurements of surface O 3 were taken at Thumba, Thiruvananthapuram (David and Nair, 2011;Girach et al., 2012), in July 2011.Along with various meteorological parameters, rainfall measurements were also taken at Thumba at 5 min of integration time using an automatic weather station manufactured by Dynalab Weathertech Pvt.Ltd, India.The site, Thumba, is situated just ∼ 500 m away from the west coast, with sandy terrain, and is a less populated area in the city of Thiruvananthapuram (8.5 • N, 76.9 • E) at southern tip of India.For more details about the Thumba site and measurements please see, for example, Nair et al. (2002) and David and Nair (2011).
A vertical profile of O 3 was measured on 28 July 2011 at Thumba using an electrochemical concentration cell ozonesonde (EN-SCI 2ZV7 ECC; Komhyr, 1969;Komhyr et al., 1995).The accuracy of such ozonesondes is reported to be about ±5-10 % up to ∼ 30 km (Smit et al., 2007).More details of this measurement technique can be found in Ojha et al. (2014).
The accumulated rainfall for every 3 h interval from the Tropical Rainfall Measuring Mission (TRMM; with a horizontal grid size of 0.25 • × 0.25 • ) is also utilised in this study to complement the on-board rainfall measurements.
The 3B42 algorithm is used to calculate precipitation and root-mean-square precipitation-error estimates; these two estimates were then used to compute hourly and daily rainfall estimates (Huffman et al., 1995).
The gridded (2 • × 2 • ) monthly column-averaged CH 4 (level 3, version 6) retrievals from SCIAMACHY (SCanning Imaging Absorption spectroMeter for Atmospheric Car-tograpHY) instrument on board the Envisat satellite were used to infer concentrations over Indian land regions.The IMAP-DOAS (iterative maximum a posteriori differential optical absorption spectroscopy) algorithm was used, which retrieves CH 4 utilising the spectra (i.e.1000-1750 nm) from the near infrared channel no.6 (Frankenberg et al., 2005).
4 Model simulations WRF-Chem (Grell et al., 2005) version 3.5.1 was used to simulate meteorological and chemical fields during the campaign period.The model domain (Fig. 2d-e) is defined on the Mercator projection, centred at 80 • E, 15.5 • N, at a spatial resolution of 15 km × 15 km.The model has 51 vertical levels from surface to 10 hPa.The simulations were conducted for the period of 29 June to 31 August 2009, covering the complete measurement period.The meteorological inputs have been adopted from ERA-Interim reanalyses by the ECMWF.Horizontal winds, temperature, and water vapour are nudged above the planetary boundary layer using a nudging coefficient of 0.0003 s −1 (Kumar et al., 2015b), employing the four-dimensional data assimilation (FDDA) technique.Anthropogenic emissions of CO, NO x , SO 2 , and NMVOCs are provided by a regional emission inventory that was developed to support INTEX-B (Zhang et al., 2009;Kumar et al., 2012b;Ojha et al., 2016).This inventory is representative of the year 2006.Aerosol emissions are provided by the Hemispheric Transport of Air Pollution (HTAP v2) inventory (Janssens-Maenhout et al., 2015).Biomass burning emissions from NCAR Fire Inventory (FINN; Wiedinmyer et al., 2011) and biogenic emissions calculated online using MEGAN (Guenther et al., 2006) were used in the simulations.
Gas-phase chemistry in the model is represented by the second-generation Regional Acid Deposition Model (RADM2; Stockwell et al., 1990), and the aerosol module is based on MADE SORGAM (Binkowski and Shankar, 1995;Ackermann et al., 1998;Schell et al., 2001).Initial and boundary conditions for chemical fields are provided by the MOZART-4/GEOS5 data.The options used to parameterise different atmospheric processes are given in Table 1.For more information about meteorological nudging, chemical mechanisms, emissions, boundary conditions, and evaluation of WRF-Chem against in situ measurements and satellite data over the Indian region, please see, for example, Kumar et al. (2012aKumar et al. ( , b, 2015a, b), b), Ansari et al. (2016), andOjha et al. (2016).Model-simulated mean spatial distributions of The mixing ratios of trace gases show large spatio-temporal variations over the BoB.Levels of O 3 and CO varied in the ranges of 8-54 (average of 29.7 ± 6.8 nmol mol −1 ) and 50-200 nmol mol −1 (average of 96 ± 25 nmol mol −1 ), respectively.As CO mixing ratios below the detection limit of the instrument are discarded from the analysis, the reported minimum and average values of CO mixing ratios are therefore slightly higher than their actual values.CH 4 mixing ratios are observed to range from 1.57 to 2.15 µmol mol −1 , with an average of 1.83 ± 0.14 nmol mol −1 .Average CH 4 mixing ratios showed a significant difference of ∼ 0.14 µmol mol −1 between northern (81-91 • E, 16-21.5 • N) and central (80-91 • E, 11-16 • N) BoB during the study period.In addition to sailing across the BoB, the ship was also kept stationary for 15 days, from 22 July to 6 August 2009  (Lin et al., 1983) graupel processes; suitable for high-resolution simulations.
Longwave radiation Rapid Radiative Transfer Accurate scheme utilises look-up tables for efficiency and Model (RRTM; Mlawer et al., 1997) accounts for multiple bands and microphysical properties.

Shortwave radiation
Goddard shortwave scheme Two-stream multi-band scheme using O 3 (Chou and Suarez, 1994) from climatology and includes cloud effects.

Surface layer
Monin-Obukhov scheme Based on Monin-Obukhov with Zilitinkevich thermal roughness (Janjic, 1996) length and standard similarity functions from look-up tables.
Land surface option Noah land surface model Unified NCEP/NCAR/AFWA scheme with soil (Chen and Dudhia, 2001) temperature and moisture in four layers, fractional snow cover and frozen soil physics.This includes the modifications for better representation of processes over ice sheets and snow covered areas.
Urban surface physics Urban canopy model Three-category urban canopy model with surface effects for roofs, walls and streets.
The HYbrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model was used to simulate 5-day backward air mass trajectories arriving at 500 m (a height that falls within the marine atmospheric boundary layer) above the measurement locations (Draxler and Rolph, 2003;Rolph, 2003; http://www.arl.noaa.gov/ready.html), as shown in the Fig. 3. Trajectories are colour-coded to show the altitude variations of the air parcels along their path.The influences of two different air masses are observed over the BoB during the CTCZ experiment.Over the central BoB, the backward air trajectories cross southern India (i.e.< 13 • N), where a belt of elevated anthropogenic emissions (5-20 mol km −2 h −1 of NO x ; see Fig. 1) is located.In contrast, most of the air trajectories over northern BoB come across the central Indian region, where anthropogenic emissions are relatively lower.For example, with the exception of a few hotspots, NO x emissions north of 13 • N are in the range of 1-10 mol km −2 h −1 (Fig. 1).The O 3 and CO mixing ratios over BoB in air masses from central/northern India (Fig. 3a) are slightly higher or comparable (O 3 : 30 ± 7; CO: 95 ± 25 nmol mol −1 ) to those (O 3 : 27 ± 5; CO: 101 ± 27 nmol mol −1 ) in air masses from southern India (Fig. 3b).
The observed spatio-temporal variations of trace gases are investigated by calculating the fractional residence time of air masses over land, using HYSPLIT simulated 5-day backward air trajectories.Figure 4a-c show the temporal variations of O 3 , CO, and CH 4 during the CTCZ experiment along the cruise track.The percentage of time of air masses over continental India is also shown (Fig. 4d), as estimated by the ratio of residence time over land to the total trajectory time of 5 days.The hours of residence have only been included in the analysis if the altitude along trajectory is less than 1.5 km, as the surface emissions might not been directly influence the air masses aloft.Red vertical bars depict the sharp reductions in O 3 mixing ratios associated with rainfall events (see Sect. 5.3).O 3 , CO, and CH 4 show correlated variability with the estimated residence times over the Indian subcontinent with slightly higher correlation in the case of primary species (R 2 = 0.16 in case of CO and CH 4 ), as compared to O 3 (R 2 = 0.09).Similar variations in mixing ratios of these trace gases and residence time over continental India indicate the influences of transport from the Indian subcontinent on the observed spatio-temporal variations over the BoB during the summer monsoon season.The occasions on which such a one-to-one correspondence are not observed can be attributed to varying source strengths, vertical mixing or dilution, and en route photochemistry.
Generally, during the summer monsoon season, relatively cleaner marine air masses from the Arabian Sea are transported to the Indian region.These air masses are then exposed to regional emissions and subjected to photochemistry depending upon the availability of solar insolation under the cloudy conditions of monsoon.The air masses in which precursors have accumulated, and to some extent photochemically processed, outflows into the BoB.As a result, the air masses outflowing at the eastern coast of India could have higher O 3 mixing ratios than the background air coming from the Arabian Sea into the western coast of India.The difference between these two values is representative of the O 3 build-up that can be attributed to regional pollution; this dif-ference would also reflect the extent of photochemical processing that would have taken place.
As the observational site Thumba, Thiruvananthapuram, is situated just at the Arabian Sea coast, the monsoon-time observations here could be approximated to represent the background O 3 mixing ratios entering from the Arabian Sea.In August 2009, using only daytime monthly average O 3 , the O 3 at Thumba during the monsoon season was observed to be 23 ± 7 nmol mol −1 .Since the objective of in-  vestigation is the additional O 3 over the BoB produced by en route photochemistry, daytime O 3 values at Thiruvananthapuram are therefore compared with all the observations over the BoB.The average mixing ratio observed over the BoB during monsoon season for 16 July-17 August 2009 was 30 ± 7 nmol mol −1 , which was ∼ 7 nmol mol −1 higher than the Arabian Sea air mass.This additional amount of ∼ 7 nmol mol −1 could be attributed to the effects of regional and en route photochemical O 3 production.Net O 3 production rate in the outflow is estimated to be in the range of 1.5-4 nmol mol −1 day −1 .Note that the O 3 mixing ratio is reported to be ∼ 30 ± 2 nmol mol −1 during July 2009 over Anantapur, a rural site in central India, indicating the enhancement due to regional O 3 production.As shown in Table 2, while average O 3 mixing ratios over the west coast of India and the Arabian Sea are in the range of 9-25 nmol mol −1 during the monsoon season, the average O 3 mixing ratio is ∼ 30 nmol mol −1 over the central Indian station and the BoB.O 3 mixing ratios were 27 ± 3 and 28 ± 5 nmol mol −1 for 21 July 2009 and 15 August 2009, for which back trajectories (not shown here) crossed Thiruvananthapuram on 20 July and 13 August 2009, with daytime O 3 values of 23 ± 6 and 25 ± 6 nmol mol −1 , respectively.The difference of 3-4 nmol mol −1 between O 3 mixing ratios over the BoB and Thiruvananthapuram represents the en route photochemical production of O 3 in the air masses toward the observation points over the BoB.It is further found that the air masses were typically below 700 m and generally within the marine boundary layer (e.g.mean boundary layer height ∼ 897 m during winter over the BoB; Subrahamanyam et al., 2012).The enhancements in O 3 are attributed here to in situ photochemical build-up while moving towards the BoB, which has been noted in previous experiments and model simulations (e.g.Lal and Lawrence, 2001;Ojha et al., 2012).
CO showed a sharp enhancement (denoted with red arrows in Fig. 4b) on 7 and 11 August 2009, coinciding with a longer residence time over the Indian region.Figure 5 shows backward air mass trajectories above the measurement locations, along with the distribution of anthropogenic CO emissions from the INTEX-B inventory, representative of the year 2006.The air masses over the BoB are found to be influenced by emission hotspots (corresponding emission of 250-350 mol km −2 h −1 ).The air masses took about half a day to be transported from the emission hotspot to the observation location over the BoB.The CO mixing ratio measured at Bhubaneswar (20.30 • N; 85.83 • E), a station within the hotspot region, is ∼ 251 ± 58 nmol mol −1 during the monsoon season (June-August 2011-2012; Mahapatra et al., 2014), with the elevated CO emissions in the Bhubaneswar region being attributed to industrial activities.The higher CO mixing ratio ∼ 200 nmol mol −1 is in line with the monsoonal values observed at Bhubaneswar.The CO mixing ratios around 150 nmol mol −1 were sampled on 11 August 2009 near the coastal source regions.Additionally, CO mixing ratios over central BoB (101 nmol mol −1 ) were only slightly higher than those over northern BoB (95 nmol mol −1 ).It is suggested that this is partially due to higher emissions over southern India against the shorter residence of air masses over land and the relatively longer lifetime of CO.
The mixing ratios of surface CH 4 were higher in the air masses from central/northern Indian over northern BoB (1.86 ± 0.12 µmol mol −1 ) as compared to those in the air masses from southern India (1.72 ± 0.14 µmol mol −1 ).As CH 4 is a relatively well-mixed trace gas, the average values over the tropospheric column approximates the uniform mixing ratio within the troposphere (Seinfeld and Pandis, 2006).The monthly column-averaged tropospheric CH 4 , retrieved from SCIAMACHY, for August 2009 shows (Fig. 6) higher values around 1.85 µmol mol −1 over central/northern India as compared to that of southern India (∼ 1.80 µmol mol −1 ).The higher tropospheric CH 4 over the central/northern Indian landmass during the summer monsoon season has been also reported by Kavitha and Nair (2016).The observed higher CH 4 over the northern BoB are attributed to the influences of emissions from central/northern Indian regions as also suggested by backward trajectories.Owing to the longer lifetime of CH 4 , diffusion of CH 4 from a hotspot re- gion over the eastern IGP to northern BoB might be the other source of higher CH 4 levels over northern BoB during summer monsoon season.An emission inventory analysis by sector over the hotspot region (i.e.eastern IGP) indicates that these higher CH 4 emissions are associated with rice cultivation, waste treatment, and livestock.The correlation between the in situ CH 4 measurements and the retrievals from the satellite instrument (AIRS -Atmospheric Infrared Sounder) was found to be statistically insignificant (not shown), which highlights a need of more such in situ measurements in this region to validate the satellite products, especially during the summer monsoon.

WRF-Chem simulations
WRF-Chem simulations, as described in Sect.4, are used to evaluate the performance of the model in reproducing the measurements and to investigate the underlying processes that caused the observed variabilities in O 3 and CO.A comparison between model-simulated and measured meteorological parameters shows only small mean biases, such as −1.9 hPa in pressure, −0.6 • C in temperature, and −1.1 % in relative humidity (Table 3).Figure 4b-c compare WRF-Chem simulated O 3 and CO with in situ measurements taken along the cruise track.WRF-Chem is found to reproduce the observed variations of O 3 and CO over the BoB during the summer monsoon season with an overestimation of absolute O 3 levels by 1.9 nmol mol −1 (i.e.∼ 6 % of averaged O 3 value, 29.7 nmol mol −1 ) and absolute CO levels by 18 nmol mol −1 (i.e.∼ 16 % of averaged CO value, 96 nmol mol −1 ).It should be noted that the average CO mixing ratio of 96 nmol mol −1 is slightly higher than its actual value, as data points below the detection limit of the instrument are discarded.Biases in the model simulations can be  (Kumar et al., 2012b).Modelled O 3 was found to be within 1σ standard deviation of mean from observations at Gadanki in the southern India during summer monsoon (Ojha et al., 2016).At Thumba also, model-simulated O 3 variations correlated reasonably with measurements (R 2 = 0.6).More information on evaluation of WRF-Chem simulations of O 3 and CO over India can be found elsewhere (e.g.Kumar et al., 2012b;Ojha et al., 2016).
An additional simulation was conducted by switching off the anthropogenic emissions over the model domain (Fig. 4b-c; dotted blue curves).Mean O 3 (17.7 nmol mol −1 ) is found to be lower by 14 with smaller variability of 2.4 nmol mol −1 compared to standard WRF-Chem simulation.Similarly, also the mean CO level is lower by 36 with smaller variability of 4.9 nmol mol −1 .This shows that enhanced levels and observed variability in O 3 and CO mixing ratios over BoB are attributable to the regional anthropogenic emissions.
The limited collocated measurements showed en route O 3 production rate of 1.5-4 nmol mol −1 day −1 , as discussed in Sect.5.1.Here, the O 3 production rate is estimated by analysing the model-simulated O 3 mixing ratios along the air mass trajectories ending at few representative locations over the BoB (Fig. 7a).The model-simulated chemical evolution of air parcels clearly shows an increase in the O 3 mixing ratios towards the marine region of BoB.The temporal variations of O 3 mixing ratio averaged corresponding all the trajectories shown in Fig. 7a is shown as a box plot in Fig. 7b.A linear regression analysis (blue curve in Fig. 7b) is used to estimate the mean en route O 3 production rate of ∼ 4.6 nmol mol −1 day −1 in the outflow.The enhancement in average O 3 over the BoB as compared to the continental Indian region is also shown in the Fig. 2d, averaged for the study period.

Ozone variations during rainfall events
An interesting phenomenon observed during the CTCZ experiment is the abrupt reduction in O 3 mixing ratios that accompanied the onset of heavy rainfall, despite the low solubility of O 3 in water.In this section the possible causes of these low-O 3 events during rainfall are investigated.(1.67-1.93)* CO mixing ratios below the detection limit (i.e.50 nmol mol −1 ) are not considered in the analysis.
Figure 8 shows variations in O 3 (black circles) mixing ratios, surface temperature (orange curve), TRMM retrieved rainfall (thick grey vertical bars), and WRF-Chem simulated vertical winds at pressure levels ranging from 950 to 750 hPa (coloured bars) during four such events on 21, 26, and 28-29 July and on 10-11 August 2009.As high time resolution in situ measurements of rainfall were not available aboard ship, Fig. 8 therefore uses 3 h rainfall retrievals from the TRMM, co-located with O 3 measurements.During these events, CO mixing ratios also show a reduction of about ∼ 56 nmol mol −1 , with observed values falling below the detection limit of the instrument during the first event of 21 July 2009 (not shown).Although CO measurements are not available for the second and third event, during the fourth event (10-11 August 2009) CO mixing ratios showed an enhancement due to transport from strong source regions ).While the first three low-O 3 events are not captured by WRF-Chem (Fig. 4c), the fourth event is reproduced.Wet scavenging does not directly reduce O 3 , as its water solubility is low; as a result, some dynamic process could be responsible for the observed reductions in O 3 during rainfall.Air masses could undergo downdrafts during heavy rainfall (Kishore Kumar et al., 2005) through air drag by the falling rain drops and in mesoscale subsidence that compensates convective updrafts.We suggest that, in the presence of O 3 -poor air mass aloft, a downdraft would result in reductions in surface O 3 mixing ratios.An opposite scenario leading to O 3 enhancement could take place if downdrafts bring mid-tropospheric air, where typically O 3 is higher than at surface.The model-simulated meteorology shows occurrences of downdrafts at different pressure levels during the first three events on 21, 26, and 28-29 July (Fig. 8a-c), which is further corroborated with measurements of air temperature aboard.Downdrafts of free-tropospheric air could lead to a reduction in near-surface temperature by as much as 10 • C within a few minutes (Ahrens, 2009).Air temperature measured aboard ship showed a sharp decrease of 2-4 • C, coinciding with the first three low-O 3 events (Fig. 8a-c).The reductions in temperature caused by downdrafts are generally short-lived (Ahrens, 2009), as is confirmed in the case of these events (Fig. 8a-c).
Model-simulated vertical winds and variations in air temperature suggest that downdrafts did occur during the first three rainfall events.As in situ measurements of O 3 vertical profiles are not available over the BoB during the summer monsoon season, the observations taken at Thumba, Thiruvananthapuram, in the southern Indian region are used as a case study to investigate this hypothesis.For general details of the typical diurnal and seasonal variations of O 3 at Thumba, please see Nair et al. (2002), David andNair (2011), andGirach et al. (2012).Figure 9a shows the temporal variation in surface O 3 on 15 July 2011 at Thumba, along with 5 min accumulated rainfall.Here, surface O 3 is observed to decline from 25 to 13 nmol mol −1 within 15-20 min, coinciding with the occurrence of intense rainfall (3.5-0.5 mm rain over a period of 5 min).Measurements of the O 3 vertical profile are not available for this day due to the rainy conditions; a profile measured on 28 July 2011 is therefore shown in Fig. 9b.This profile has lower O 3 mixing ratios aloft (∼ 22 nmol mol −1 at ∼ 1 km) than near the surface (∼ 42 nmol mol −1 ).The observed mixed layer height is about 0.15-0.6km over Thumba, Thiruvananthapuram (Anurose et al., 2016), during July 2011; above this height, O 3 mixing ratios sharply decrease with altitude.The present case study suggests the presence of O 3 -poor air masses aloft than those near the surface over the south Indian region during summer monsoon.With an O 3 distribution as observed in the present case study at Thumba, the downdraft during intense rainfall could lead to the mixing of free-tropospheric 1 2 : 4 5 1 3 : 0 5 1 3 : 2 5 1 3 : 4 5 1 4 : 0 5 1 4 : 2 5 1 4 : 4 5  1  air with near-surface air or to the replacement of surface air with free-tropospheric O 3 -poor air.
Although air temperature measurements could not be made during the fourth event (10-11 August 2009) due to a technical problem, model meteorology does not indicate a downdraft during this event (Fig. 8d), indicating the dominance of a different process.As WRF-Chem simulated O 3 variability is in good agreement with observations during this event, various tendency terms (Barth et al., 2012) from WRF-Chem are used to investigate the relative influences of different processes.The variations in instantaneous values for horizontal advection tendency, vertical advection tendency, and net tendency (i.e. the sum of chemical, vertical mixing, convective, vertical advection, and horizontal advection), along with modelled O 3 over the two locations during the event, are shown in Fig. 10.The tendency values shown here are derived by subtracting the accumulated tendencies at (n − 1)th hour from the accumulated tendency at nth hour.The vertical dotted lines show the time of a low-O 3 event.
Both the horizontal and net tendencies (Fig. 10b-c, e-f) show negative values, indicating that they are contributing towards a reduction in O 3 mixing ratios (Fig. 10a, d).However, as the time of the event approaches, it is the horizontal advection tendency term that is significantly negative (Fig. 10c, f), while other terms are small and close to zero.Horizontal advection is therefore suggested to dominate during the low-O 3 event of 10-11 August 2009.The influence of horizon- tal advection on O 3 during this event is shown more clearly in Fig. 11, which shows the spatial distribution of O 3 and CO from WRF-Chem before the event (16:00 and 19:00 UT) and during the event (22:00 UT) on 10 August 2009.The white triangles show the two locations where the event was observed.During 16:00 and 19:00 UT, a patch of high O 3 mixing ratios (35 nmol mol −1 and higher) is seen to be distributed over a large region surrounding the measurement location.This large patch of elevated O 3 mixing ratios is horizontally advected eastward from 16:00 to 19:00 and then towards 22:00 UT (event time).As a result of this rapid advection, the high-O 3 air masses are transported from the coastal regions to deeper into the BoB; by the time they reached the location of observation, O 3 mixing ratios are observed to be lower (25-35 nmol mol −1 ) during the event time (22:00 UT).A patch of higher levels of CO (∼ 300 nmol mol −1 ) was also found to be distributed across the east coast of the Indian region.Transport and dilution of this CO patch is, however, less pronounced than the high-O 3 air masses, possibly due to the relatively longer lifetime of CO.Thus, in a nutshell, the horizontal advection played a key role in transporting O 3 -rich air masses deeper into the BoB region, while it diluted O 3 levels near the coastal regions in southern India during the fourth event.

Seasonal variation in trace gases over the BoB
In this section, the monsoon-time measurements of O 3 taken in the present study are combined with data from previous campaigns (see Table 4) to investigate the seasonal variation in O 3 over the BoB (Fig. 12).O 3 , CO, and CH 4 mixing ratios are averaged over northern (81-91 • E, 16-21.5 • N) and central (80-91 • E, 11-16 • N) BoB regions.
Overall, higher O 3 mixing ratios are present over both northern and central BoB during the winter, while lower O 3 levels are observed during the spring-summer (with more scatter in the data over central BoB).The O 3 seasonal amplitude (i.e. the range from maxima to minima) is estimated to be ∼ 39 over northern BoB and ∼ 27 nmol mol −1 over central BoB.The monsoonal surface O 3 mixing ratios (∼ 30 ± 7 nmol mol −1 ) are nearly half those observed during winter (63 ± 5 nmol mol −1 ) over northern BoB. During December 2008-January 2009, February 2003, March 2006, and November 2010, the O 3 mixing ratios were higher (by ∼ 3-22 nmol mol −1 ) over northern BoB than over central BoB.However, over the course of February 2001, O 3 mixing ratios were higher over central BoB (∼ 38 nmol mol −1 ) than over northern BoB (∼ 14 nmol mol −1 ).In contrast, during summer monsoon season, average O 3 mixing ratios are comparable or only slightly higher over northern BoB (30 ± 7 nmol mol −1 ) as compared to over central BoB (27 ± 5 nmol mol −1 ).
As compared with the summer monsoon season, when CO mixing ratios were lower over northern BoB, CO mixing ratios were higher during the winter, while over central BoB CO mixing ratios were higher during the pre-monsoon season.For O 3 , spring-summer had the lower mixing ratios in both regions.The seasonal amplitude in CO mixing ratios is estimated to be ∼ 205 over northern BoB and ∼ 124 nmol mol −1 over central BoB.The monsoonal CO mixing ratio (∼ 95 nmol mol −1 ) is about one-third that of the winter season (302 nmol mol −1 ) over northern BoB.During the present study, average CO mixing ratios were comparable over northern (95 ± 25 nmol mol −1 ) and central BoB (101 ± 27 nmol mol −1 ).
A clear inference about seasonal patterns is difficult in the case of CH 4 , but a tendency of lower levels towards winter can be seen.Higher mixing ratios ∼ 1.95 (∼ 1.91) µmol mol −1 were observed during November 2010 over northern BoB, and during February-March 2001 over central BoB, as compared to those from other studies.The surface CH 4 observations obtained during the present study show the highest variability (i.e. the difference between maxima and minima) when compared to earlier studies: 0.53 over northern BoB and 0.39 µmol mol −1 over central BoB.This high variability is attributed to the relative source strengths over central and northern India as compared to southern India, highlighting the regional differences in CH 4 variability across India (Kavitha and Nair, 2016).(David et al., 2011;Lal et al., 2007Lal et al., , 2006;;Nair et al., 2011;Srivastava et al., 2012;Sahu et al., 2006;and Mallik et al., 2013).Error bars show standard deviations for respective study periods.For any points for which high-resolution measurements are not available, standard deviations are not shown.
Seasonal variations in trace gases over the BoB are attributed to seasonal changes in the meteorological conditions, emissions, and photochemistry over the South Asian region, as well as to synoptic-scale transport patterns.Wintertime stronger westerly winds transport the pollution from South Asia, including that of the Indo-Gangetic basin, to the BoB region.Monsoonal circulation, in contrast, carries cleaner marine air masses to the BoB from the oceanic regions.However, as observed during the CTCZ, polluted continental or coastal air masses can also occasionally be transported deeper over the BoB.Intense monsoonal rainfall generally leads to wet removal of O 3 precursors, while cloudy and rainy meteorological conditions suppress O 3 formation.Along with the importance of monsoonal convection in cloud formation, rainfall, and uplifting the boundary layer pollution, rapid horizontal advection is also an important process during the summer monsoon, especially affecting the nearsurface variability of trace gases over the oceanic regions adjacent to India.

Conclusions
In this paper, we presented shipborne in situ measurements of O 3 , CO, and CH 4 that were carried out as a part of the CTCZ experiment over the BoB during the summer monsoon season, July-August 2009.We analysed the spatial and temporal variations in the observations and compared them with results from simulations performed with a regional chemistry transport model (WRF-Chem), as well as with observations from previous campaigns over the BoB.The main conclusions are as follows.
Back-trajectory analysis shows effects of long-range transport from northern or central India to northern BoB and from southern India to central BoB.The correlated variations of these trace gases and percentage residence time of air parcels over the Indian regions suggest that the enrichment of O 3 and precursors in air parcels over the BoB is associated with both emissions and photochemistry over the Indian region.The trajectory analysis shows that the observed variation in surface O 3 is primarily due to transport and en route photochemistry over the BoB during monsoon season.An analysis of modelled O 3 along air mass trajectories show mean en route O 3 production rate of about 4.6 nmol mol −1 day −1 in the outflow towards the BoB.
The observed spatio-temporal variations of surface O 3 and CO during summer monsoon season are generally reproduced by WRF-Chem, although the absolute mixing ratios of O 3 and CO are typically overestimated by about 6 and 16 %, respectively.
The four low-O 3 events coinciding with intense rainfall were observed over the BoB.After analysing the observed variability in air temperature, model simulations of vertical winds, and an O 3 -profile case study from southern India, we suggest that first three low-O 3 events were due to strong www.atmos-chem-phys.net/17/257/2017/Atmos.Chem.Phys., 17, 257-275, 2017 downdrafts of O 3 -poor air masses.Analysis of the fourth event, which is successfully reproduced by the model, shows the pivotal role of horizontal advection in transporting O 3rich air masses deeper over the BoB.Finally, the measurements during the monsoon are combined with previous campaigns over the BoB during other seasons to investigate the seasonal variability in trace gases over the BoB.O 3 and CO are shown to have pronounced seasonality, O 3 having amplitudes of about 39 and 27 nmol mol −1 and CO having amplitudes of about 207 and 124 nmol mol −1 over northern and central BoB, respectively.
Our study data fill a gap of observations during the summer monsoon over the BoB, providing information on the extent of seasonal variability.We recommend supplementing these findings with shipborne experiments featuring collocated vertical profile observations from balloon-borne and aircraft-based platforms over the oceanic regions surrounding India to better understand the role of both large-scale dynamics (e.g.Ojha et al., 2016) and of regional influences due to South Asian outflow (see Lawrence and Lelieveld, 2010, and references therein).Such a future study would also improve our understanding of the changes that take place in the atmospheric oxidation capacity during the summer monsoon season.

Data availability
The observational data used in the paper can be obtained by contacting the corresponding author (imran.girach@gmail.com).The metadata are available on the website of Space Physics Laboratory (http://spl.gov.in/index.php?option=com_content&view= article&id=394&Itemid=695&lang=en).

Figure 1 .
Figure 1.Cruise track (continuous black line) of the Research Vessel Sagar Kanya along with synoptic winds at 925 hPa (black thin arrows) and NO x emissions in background colour map.The dates corresponding to approximate ship positions are marked along the track.The start and end position of the cruise, stationary position of the ship, and Thiruvananthapuram are shown by the circle, triangle and square, respectively.

Figure 2 .
Figure 2. Spatial variation of surface O 3 (a), CO (b), and CH 4 (c) mixing ratios along the cruise track during the CTCZ campaign.WRF-Chem simulated spatial distribution of surface O 3 (d) and CO (e) averaged during 16 July-17 August 2009.

Figure 3 .
Figure 3. Five-day air mass back trajectories during the study period ending at the measurement locations (small black circles) grouped for corresponding air masses from (a) central/northern India and (b) southern India.The cross symbols along the trajectories represent each back-day.The colour scale shows the height (in km) of the trajectories.

Figure 4 .
Figure 4. Variations in observed CH 4 (a), CO (b), O 3 (c), and percentage residence time over land (d) along with WRF-Chem simulated O 3 and CO (blue line) during the campaign.Events of sharp decrease in O 3 during rainfall are marked by vertical red lines (Fig. 8).(e) Variations in measurement locations.

Figure 5 .
Figure 5. Backward air mass trajectories (black curves) 500 m above the location of higher CO observations as marked by red arrows in Fig. 4b during 7 and 11 August 2009.The background colour map shows the spatial distribution of anthropogenic CO emissions over the Indian region for the year 2006 from INTEX-B inventory.The small circles in magenta represent the points where observations were made, as well as the end point of trajectories.The white circle over the hotspot region denotes an observational site, Bhubaneswar.

Figure 6 .
Figure 6.Spatial distribution of column-averaged CH 4 for the month of August 2009 as obtained from SCIAMACHY.

Figure 7 .
Figure 7. (a) WRF-Chem simulated O 3 along the air mass trajectories ending over a few representative locations over the BoB.(b) Variation of O 3 mixing ratios with time along the trajectories as shown in (a).In box plot, the black dots and lines inside the box represent the mean and median of the data, respectively.While the lower and upper edges of boxes represent the 25th and 75th percentiles, respectively, the whiskers represent standard deviations.

Figure 8 .
Figure 8. Surface O 3 (black dots) along with temperature (orange curve) and 3 h rainfall (grey vertical bar) during the four events of sharp decline in O 3 (a-d) as marked in Fig. 4c.Colours indicate the vertical wind as simulated by WRF-Chem.
Figure 9. (a) Temporal variation in surface O 3 mixing ratio (black dots) along with 5 min accumulated rainfall (grey vertical bars) over Thumba, Thiruvananthapuram (location of the site shown in Fig. 1), on 15 July 2011.(b) Vertical profile of O 3 mixing ratio over Thumba, Thiruvananthapuram, as measured on 28 July 2011.

Figure 10 .
Figure 10.Time series of surface O 3 (a) and various tendency terms (b and c) over the event location during the fourth low-O 3 event, as obtained from WRF-Chem simulations.(d)-(f) are the same as (a)-(c), but for another location during the same event.These two event locations are also marked by triangles in Fig. 11.Vertical dotted line shows the time of the event in the in situ observations of surface O 3 over the indicated locations.

Figure 11 .
Figure 11.Spatial distribution of surface O 3 (top panel) and CO (bottom panel) at 16:00 and 19:00 UT on 10 August 2009, both prior to and during the fourth event, which took place 22:00 UT on 10 August 2009.White triangles show two locations (85.40 • E, 19.00 • N; 84.95 • E, 19.00 • N) corresponding to the event.

Figure 12 .
Figure 12.Seasonal variation in average O 3 , CO, and CH 4 mixing ratios over (a, c, e) northern BoB and (b, d, f) central BoB.Except for July-August 2009 (present study period), all average values are obtained from the literature(David et al., 2011;Lal et al., 2007Lal et al., , 2006;;Nair et al., 2011;Srivastava et al., 2012;Sahu et al., 2006;and Mallik et al., 2013).Error bars show standard deviations for respective study periods.For any points for which high-resolution measurements are not available, standard deviations are not shown.
3 • E, 13.1 • N; marked by a circle) on 16 July 2009.The cruise offered greater coverage in the northern BoB than the southern or central BoB areas.To take time series measurements, the ship was kept stationary for 15 days (22 July to 6 August 2009) at 89 • E, 19 • N as marked by a triangle in the figure.After several tracks, covering latitude sector 11.0 to 21.1 • N and longitude sector 80.3 to 90.1 • E, the cruise ended on 17 August 2009 at Chennai, for a total of 32 days of voyage.

Table 1 .
The WRF-Chem options used for parameterisation of atmospheric processes.

Table 2 .
A comparison of averaged surface O 3 mixing ratios measured at various sites during summer monsoon period.
* Boundary layer O 3 over the Arabian Sea.

Table 3 .
A comparison of mean values from observations with model-simulated parameters along with the mean bias.The squared correlation coefficients correspond to the linear regression analysis between daily averaged in situ and simulated parameters.The higher value of the squared correlation coefficient for O 3 demonstrates WRF-Chem's ability to reproduce the observed broad features in surface O 3 over the BoB.Note that the sharp reductions that caused very low O 3 during rainfall episodes are not captured by WRF-Chem, except during the event of 10-11 August 2009.This will be discussed in detail in Sect.5.3.WRF-Chem simulated O 3 has also been evaluated against several surface observations in India previously

Table 4 .
A comparison of average mixing ratios of surface trace gases measured over northern BoB (81-91 • E, 16-21.5 • N) and central BoB (80-91 • E, 11-16 • N) in different seasons as measured during different experiments.The range of mixing ratios (i.e.minima-maxima) is given in the brackets.