Ozone and carbon monoxide over India during the summer monsoon: regional emissions and transport

We compare in situ measurements of ozone ( O 3 ) and carbon ( CO ) proﬁles from the CARIBIC program with the results from the regional chemistry transport model (WRF-Chem) to investigate the role of local/regional emissions and long-range transport over southern India during the summer monsoon of 2008. WRF-Chem successfully reproduces 5 the general features of O 3 and CO distributions over the South Asian region. However, absolute CO concentrations in the lower troposphere are typically underestimated. Here we investigate the inﬂuence of local relative to remote emissions through sensitivity simulations. The inﬂuence of 50 % increased CO emissions over South Asia leads to a signiﬁcant enhancement (upto 20 % in July) in upper tropospheric CO in the northern and central 10 Indian regions. Over Chennai in southern India, this causes a 33 % increase in surface CO during June. However, the inﬂuence of enhanced local /regional emissions is found to be smaller (5 % ) in the free troposphere over Chennai, except during September. Local to regional emissions are therefore suggested to play a minor role in the underestimation of CO by WRF-Chem during June–August. In the lower troposphere, a high pollution ( O 3 : 15 146 . 4 ± 12 . 8 nmol mol − 1 , CO : 136 . 4 ± 12 . 2 nmol mol − 1 ) event (15 July 2008), not reproduced by the model, is shown to be due to transport of photochemically processed air masses from the boundary layer in southern India. A sensitivity simulation combined with backward trajectories indicates that long-range transport of CO to southern India is signiﬁcantly underestimated, particularly in air masses from the west, i.e. from Central Africa. This study 20 highlights the need for more aircraft-based measurements over India and adjacent regions and the improvement of global emission inventories. CARIBIC measurements of ozone and CO proﬁles conducted during in conjunction with the regional chemistry transport model (WRF-Chem) to assess the contributions of emissions and long-range transport over the southern region. WRF-chem simulations are evaluated against the in situ CARIBIC proﬁles, an ozonesonde climatology, satellite (MOPITT) retrievals and ground-based measurements to identify the strengths and limitations of the WRF-Chem simulations. Additionally, we conduct a set of sensitivity to identify the role of anthropogenic emissions and long-range transport/boundary conditions.


Introduction
Tropospheric ozone and its precursors play vital roles in atmospheric chemistry, air quality degradation and climate change (e.g. WHO, 2003;Stevenson et al., 2013;Monks et al., 25 2014). It is therefore important to understand the spatial and temporal distributions of these Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | species and the contributions of different sources to their atmospheric budgets. Additionally, relative contributions of regional anthropogenic emissions and long-range transport need to be addressed for adequate policy making. In order to understand the quantitative contributions of different sources and processes (chemistry, transport) to the budgets of trace gases, systematic measurements of the vertical distribution of trace species are required in 5 conjunction with chemistry-transport modeling.
Unfortunately, in situ measurements of vertical distribution of ozone and related trace gases are very sparse over the South Asian region, where rapidly increasing anthropogenic emissions lead to severe air pollution in recent years (e.g. Akimoto, 2003;Beig and Brasseur, 2006;Ohara et al., 2007;Lelieveld et al., 2013;Pozzer et al., 2015). It is 10 suggested that air quality will further deteriorate to become severe over India in a Businessas-Usual (BAU) scenario (Pozzer et al., 2012). A recent study shows that ozone pollution alone could lead to a loss of crop yield which could feed 94 million people below the poverty threshold in India (Ghude et al., 2014). Observations such as the Indian Ocean Experiment (INDOEX) (Lelieveld et al., 2001) and the Integrated Campaign for Aerosols, Gases and 15 Radiation Budget (ICARB) (Srivastava et al., 2011) have revealed significant South Asian outflow over the surrounding marine regions (Lawrence and Lelieveld, 2010, and references therein). Additionally, the emissions and photochemically processed air masses can be uplifted due to strong tropical convection and can be transported to distant regions (Lelieveld et al., 2002;Lawrence et al., 2003;Park et al., 2007) influencing global air quality and 20 climate.
Numerous efforts have been initiated to conduct in situ ground-based measurements of ozone and precursors (e.g. Lal et al., 2000;Reddy et al., 2008;David and Nair, 2011;Sarangi et al., 2014). and ship-based (e.g. Sahu and Lal, 2006;Mallik et al., 2013;Lawrence and Lelieveld, 2010, and references therein). However, most of the studies over this region 25 have been confined to the surface. The observational studies were followed by utilizing global chemistry-transport models, such as MATCH-MPIC Ojha et al., 2012), MOZART (Beig and Brasseur, 2006;Sheel et al., 2010), and recently with a regional chemistry-transport model (WRF-Chem) Michael et al., 2014). Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | WRF-Chem simulations were generally found to reproduce the variations observed in ground-based and ozonesonde measurements over India . Model evaluation over an urban site in the Indo-Gangetic plain (Michael et al., 2014) showed an increase in model biases in simulating O 3 and CO towards the onset of monsoon as compared to spring. Model results were also evaluated against satellite retrievals of NO 2 and CO (Ku- 5 mar et al., 2012). These studies suggested that WRF-Chem at higher resolution could better capture the variations in trace gases and aerosols than global models over the Indian region because of better dealing with the complex topography and large spatio-temporal heterogeneity in the emissions. However, evaluation of WRF-Chem simulations over the Indian region is still very limited, particularly against in situ measurements of vertical profiles 10  .
The studies over the Indian region utilizing the WRF-Chem model have revealed significant differences between the model simulations and measurements, which have been attributed mainly to uncertainties in anthropogenic emissions Michael et al., 2014). Transport of CO has been investigated for the winter season by evaluating 15 the model against satellite datasets (Kumar et al., 2013) in the absence of in situ observations of vertical profiles. Lack of in situ measurements in the free troposphere and above has inhibited the quantitative understanding of the transport involved, which could play a significant role in the free troposphere (e.g. Lal et al., 2013Lal et al., , 2014Ojha et al., 2014). CARIBIC (Civil Aircraft for the Regular Investigation of the Atmosphere Based on an Instru- 20 ment Container) observations (http://www.caribic-atmospheric.com/, Brenninkmeijer et al., 2007;Rauthe-Schöch et al., 2015) can partly fill this gap by providing in situ measurements of ozone and CO profiles over the Indian region.
The Asian summer monsoon is a dominant atmospheric phenomenon over the Indian region and is shown to redistribute trace gases and aerosols (Park et al., 2009;Randel 25 et al., 2010;Baker et al., 2011;Cristofanelli et al., 2014;Fadnavis et al., 2013Fadnavis et al., , 2015. The monsoonal convection uplifts and mixes regional pollution into the upper troposphere, while the anticyclonic winds can bring polluted air masses from other regions and also export the monsoon air with its pollution to regions far away from India. However, the influences Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | of local and regional emissions compared to long-range transport are not well understood, primarily due to lack of in situ measurements of vertical profiles for the evaluation of model simulations. A few studies have utilized satellite datasets, however the view of satellite instruments during the monsoon period is often obscured by clouds. While, global models have the advantage of including large-scale dynamics /long-range transport, the regional 5 models offer better opportunity to investigate the effects of high-resolution regional emissions and regional photochemistry. The inflow of pollution to the regional models is generally provided in form of time-varying chemical boundary conditions from global model simulations (e.g. Pfister et al., 2013;Andersson et al., 2015). Therefore, long-range pollution transport is accounted for but at reduced time resolution compared to global models. 10 This study utilizes CARIBIC measurements of ozone and CO profiles conducted during the summer monsoon period (June to September) in the year 2008 in conjunction with the regional chemistry transport model (WRF-Chem) to assess the contributions of emissions and long-range transport over the southern Indian region. WRF-chem simulations are evaluated against the in situ CARIBIC profiles, an ozonesonde climatology, satellite (MOPITT) 15 retrievals and ground-based measurements to identify the strengths and limitations of the WRF-Chem simulations. Additionally, we conduct a set of sensitivity simulations to identify the role of anthropogenic emissions and long-range transport/boundary conditions. The paper is structured as follows: the configuration of the WRF-Chem model used in the present study is described in Sect. 2. The CARIBIC measurements, satellite data 20 and ground-based measurements used to evaluate the model simulations are discussed in Sect. 3. The results with a focus on model evaluation are presented in Sect. 4.1, followed by an investigation of the influences of regional emissions (Sect. 4.2) and transport (Sect. 4.3). The summary and conclusions are presented in Sect. 5.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 2 Model description and setup

WRF-Chem
We use version 3.5.1 of the Weather Research and Forecasting with Chemistry (WRF-Chem), an online regional chemistry transport model (Grell et al., 2005). The simulation domain has been defined on the Mercator projection (Fig. 1). The model domain is centered 5 at 80 • E, 22 • N, and covers nearly the entire South Asian region with a spatial resolution of 30 km × 30 km. In the west-east and south-north directions, the domain has 132 and 120 grid points. We have used 51 vertical levels in the model starting from surface to 10 hPa.
The geographical data e.g. terrain height, land-use etc. have been interpolated from the USGS (United States Geological Survey, Wang et al., 2014) data at 10 min resolution for 10 the model domain using the geogrid program of the WRF Preprocessing System (WPS). The different options used in this study to parametrize the atmospheric processes are listed in Table 1. The instantaneous model output has been stored every hour and has been used for the analysis. Model simulations are conducted for the period of 29 May to 30 September 2008. First three days output was discarded as the model spin up. 15 NCEP Final Analysis (FNL from GFS ds083.2) dataset (http://rda.ucar.edu/datasets/ ds083.2/) with a spatial resolution of 1 • , available every 6 h, has been used to provide the initial and lateral boundary conditions for the meteorological fields. Four Dimensional Data Assimilation (FDDA) has been applied to limit the errors in the simulations of meteorological parameters. The horizontal winds, temperature and water vapor are nudged with a nudging 20 coefficient of 0.0006 at all vertical levels. The time step for the simulations has been set at 120 s, which is 4 times the grid resolution (30 km), so that the CFL stability criterion is not violated.
The anthropogenic emissions of CO, NO x , SO 2 , NMVOCs, PM, BC and OC are from the Hemispheric Transport of Air Pollution (HTAP v2) emission inventory available on a monthly 25 temporal resolution (http://edgar.jrc.ec.europa.eu/htap_v2/index.php?SECURE=123). The HTAP emissions are based upon compilation of regional emission inventories available from US EPA for USA, Environment Canada for Canada, EMEP and TNO for Europe and MICS-Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Asia for Asian countries including India. The rest of the world is filled by emissions from EDGAR4.3. The HTAP v2 data are harmonized at a spatial resolution of 0.1 • × 0.1 • for the years 2008 and 2010. We have utilized the emissions available for the year 2008. The emissions available from different sectors such as energy, industry, residential, ground-transport, ships and agriculture have been combined and then mapped on the WRF-Chem grid.  tailed information on the HTAP inventory used can be found in a recent study (Janssens-Maenhout et al., 2015). An additional simulation has also been conducted using a different regional inventory (INTEX-B) (Zhang et al., 2009) for anthropogenic emissions.
The biomass burning emissions to the model have been provided from the Fire INventory from NCAR (FINN), Version 1 (Wiedinmyer et al., 2011). The biogenic emissions are 10 calculated using the Model of Emissions of Gases and Aerosols from Nature (MEGAN) (Guenther et al., 2006) online based on weather and land use data. The gas phase chemistry is represented by the second generation Regional Acid deposition Model (RADM2) (Stockwell et al., 1990), which includes 63 chemical species participating in 21 photolysis and 136 gas phase reactions. The aerosol module is based on the Modal Aerosol Dynamics 15 Model for Europe (MADE) (Binkowski and Shankar, 1995;Ackermann et al., 1998) and Secondary Organic Aerosol Model (SORGAM) (Schell et al., 2001). GOCART dust emissions have been included with AFWA modifications. The feedback from aerosols to the radiation scheme has been turned on in the simulations.
The spatial resolution of MOZART-4/GEOS5 simulations is 1.9 • × 2.5 • and has 56 pres-sure levels from surface to about 2 hPa. In this study we show the simulations driven by MOZART-4/GEOS5 initial and boundary conditions. A sensitivity analysis (not shown here) using MOZART4/NCEP data revealed similar results for WRF-Chem simulated free tropospheric ozone and carbon monoxide. We have conducted three different WRF-Chem simulations by varying the initial and 5 boundary conditions and the anthropogenic emissions as mentioned in Table 2. The first simulation called as "Std" is the standard run without any adjustments in the anthropogenic emissions or MOZART-4/GEOS5 boundary conditions data (Sect. 4.1.1). The additional simulation 1.5×_EM has been conducted by enhancing the anthropogenic emissions over South Asia by a factor of 1.5 to investigate the influence of regional emissions (Sect. 4.2). 10 The simulation 1.25×_BDY has been conducted by enhancing the CO mixing ratios by a factor of 1.25 on the MOZART boundary conditions data at the western fringe of the domain (Sect. 4.3) to study the effect of long-range transport.

Backward trajectories
In order to investigate the transport of CO  Hess, 1997, 1998;Draxler et al., 2014) and use of other meteorological datasets as inputs to the HYSPLIT model over the Indian region can be found elsewhere (Ojha et al., 2012;Sarangi et al., 2014). 25 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 3 Observational datasets

CARIBIC
This study primarily utilizes the in situ measurements of ozone and CO vertical profiles collected over Chennai in the southern Indian region as a part of the CARIBIC project. The CARIBIC observatory is deployed on a monthly basis aboard a Lufthansa Airbus A340-600 5 for a series of two to six long distance flights. The aircraft is fitted with a permanently mounted inlet system which is connected via stainless steel tubing to the CARIBIC instrument container when installed. Parts of the tubing are lined with thin walled PFA tubes to avoid wall effects (Brenninkmeijer et al., 2007). From April to December 2008, the CARIBIC container measured during 32 flights between Frankfurt, Germany and Chennai. Here we CO is measured with a commercial AeroLaser AL 5002 resonance fluorescence UV instrument modified for use onboard the CARIBIC passenger aircraft. Alterations were necessary to optimise the instrument reliability to allow for automated operation over an entire CARIBIC flight sequence lasting several days. The instrument has a precision of better than 20 2 nmol mol −1 at an integration time of 1 s. For more details, the reader is referred to Scharffe et al. (2012).
The ozone measurements are performed by a fast, commercially available dry chemiluminescence instrument, which at typical ozone mixing ratios between 10 and 100 nmol mol −1 and a measurement frequency of 10 Hz has a precision of better than 1.0 %. The absolute 25 ozone concentration is inferred from a UV-photometer designed in-house which operates at 0.25 Hz and reaches an accuracy of 0.5 nmol mol −1 . More technical details can be found in Zahn et al. (2012). Water vapor mixing ratios were measured using a modified two-channel Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | photo-acoustic diode-laser spectrometer with a precision of 1 ppmv. These measurements were calibrated using the frost-point hygrometer (Zahn et al., 2014;Rauthe-Schöch et al., 2015).

Balloon-borne Measurements
WRF-Chem simulations have also been compared with the ozonesonde observations at The ozonesondes apply a modified electrochemical ozone sensor (Shreedharan , 1968). 10 These ozonesondes have also been a part of the JOSIE intercomparison experiment (Smit and Kley, 1998) . The ozonesonde observations have been previously used for analysis of long-term changes in tropospheric ozone over India (Saraf and Beig, 2004). Considering the very low temporal frequency of these observations (lack of any profiles over Delhi during the year 2008 (as of CARIBIC) and lack of observations in individual months 15 e.g. in September over Pune; this dataset has been converted to a monsoon time (June-September) climatology for 2006-2009 period around 2008 for comparison with WRF-Chem simulations. Day-to-day variability in model simulated wind speed at different pressure levels has also been evaluated above Chennai against the radiosonde observations available at http://weather.uwyo.edu/upperair/sounding.html. 20

Satellite data
We also use vertical profiles of CO retrieved from the Measurements of Pollution in the Troposphere (MOPITT) instrument (https://www2.acd.ucar.edu/mopitt) for comparison with WRF-Chem simulations. The MOPITT instrument on the EOS-Terra provides the vertical profiles and global distribution of tropospheric CO with the expected precision of 10 % 25 (Pan et al., 1998;Deeter et al., 2003;Yoon et al., 2013). Because MOPITT measures up-Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | welling infrared radiation at 4.7 and 2.4 µm, it can provide data during night and day. Even though the retrieval sensitivity is generally greater for daytime than for nighttime overpasses (Deeter, 2013), the nighttime retrievals have been updated by using the improved a priori profiles over land (Ho et al., 2005).
We have used the gridded monthly CO retrievals (MOP03JM) version 6 data, available at 5 http://reverb.echo.nasa.gov/reverb/. The major updates with this version include corrected geolocation data, use of NASA MERRA reanalysis product for meteorological fields and a priori surface skin temperatures instead of NCEP and updated CO a priori (Deeter, 2013). The Thermal and Near IR (JIR) retrievals have been utilized to have better sensitivity of MOPITT retrievals in the lower free tropospheric altitudes (Worden et al., 2010). Further 10 information on MOPITT CO retrievals (Deeter et al., 2003(Deeter et al., , 2004a and comparison with measurements (Emmons et al., 2004(Emmons et al., , 2007 and model simulations can be found elsewhere (Emmons et al., 2010;Yoon and Pozzer, 2014).

Ground-based measurements
Ground-based measurements of CO used in the study are obtained from Cape Rama Ground-based measurements of ozone at the rural site Gadanki (79.2 • E, 13.5 • N) were 25 obtained from the literature (Naja and Lal, 2002;Renuka et al., 2014). These measurements are based upon the UV-absorption technique. The accuracy of these instruments is reported to be ±5 % (Kleinman et al., 1994).
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 4 Results and discussion

Model evaluation
In this section, WRF-Chem simulated ozone and carbon monoxide data over Chennai are evaluated against the CARIBIC observations, MOPITT retrievals of CO profiles and the ground-based measurements.

Comparison with CARIBIC profiles
The hourly output of WRF-Chem simulations have been spatially and temporally interpolated along the CARIBIC flight tracks. The observed and model simulated profiles have been averaged into vertical bins of 50 hPa for the comparison analysis. The comparison of O 3 and CO profiles from CARIBIC measurements with standard WRF-Chem simulations is 10 shown in Fig. 2. Here we only show the profiles collected during the descent of the aircraft as these have complete coverage until about 800 hPa, while the measurements start from about 600 hPa onwards in the ascending profiles. However for the analysis of model biases, all the ascending and descending profiles have been averaged to calculate the monthly profiles (Fig. 3). 15 Higher levels of ozone and carbon monoxide occur in the Lower Troposphere (LT: 850-600 hPa) and Upper Troposphere (UT: above 300 hPa), while lower levels in the Middle Troposphere (MT: 600-400 hPa), cause a typical C-shape structure during July. This feature is suggested to be associated with the monsoonal convective uplifting of the lower tropospheric pollution and is captured by WRF-Chem. 20 Despite the qualitative agreement of the vertical distributions of O 3 and CO, significant differences occur between model and measurements, particularly in lower tropospheric CO. For example, on 19 June the observational CO levels vary from 91.5 ± 3.9 to 104.4 ± 0.6 nmol mol −1 in the LT, whereas WRF-Chem simulated CO levels are significantly lower (75.4±1.0 to 85.8±0.7 nmol mol −1 ). The average underestimation (Mean Bias) of CO 25 in the LT is found to be 12.6 ± 4.4, 22.8 ± 12.6 and 19.9 ± 7.5 nmol mol −1 during June, July Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | and August respectively, as calculated from all the ascend and descend profiles averaged for a month (Fig. 3). WRF-Chem simulated average CO shows very good agreement with CARIBIC measurements during September in the LT (MB = −0.1 ± 4.2 nmol mol −1 ).
The model underestimates a pollution event of strongly elevated ozone observed on 15 July 2008 (146.4 ± 12.8 nmol mol −1 at 810 hPa). In contrast to CO which is typically 5 underestimated in LT, the bias in model simulated O 3 varies from an overestimation by 4.3 ± 1.8 during June to an underestimation by 7.8 ± 1.6 nmol mol −1 during August, except during the strong pollution event (−71.5 ± 25.9 nmol mol −1 ). The significantly higher levels of O 3 (146.4 ± 12.8 nmol mol −1 ) and CO (136.4 ± 12.2 nmol mol −1 ) as observed during July are from two observational profiles on the same day (15 July

Comparison with Ozonesonde Climatology
WRF-chem simulated ozone profiles are compared with the monsoon-time climatology obtained from ozonesonde observations at Delhi, Pune and Thiruvananthapuram (Fig. 5), as described in subsection 3.2. WRF-Chem simulated ozone profiles in the lower and middle troposphere are generally observed to be within the 1-standard deviation variability of 25 the observational climatology over the three stations. However, in the upper troposphere, WRF-Chem overestimates ozone mixing ratios over Delhi and Pune. The mean biases of the WRF-Chem are estimated against average ozonesonde climatology in summer mon-MB in the LT are found to be lower at Delhi (−2.2 ± 3.8 nmol mol −1 ) and Pune (−1.2 ± 3.6 nmol mol −1 ), as compared to that over Thiruvananthapuram (−12.4 ± 1.3 nmol mol −1 ).
However, in the UT (e.g. at 150 hPa) ozone mixing ratios in WRF-Chem simulations at Delhi (94.1 ± 31.1 nmol mol −1 ) and Pune (69.4 ± 23.5 nmol mol −1 ) are found to be higher 5 as compared to ozonesonde observations (61.1 ± 34.0 and 31.3 ± 17.5 nmol mol −1 respectively). The overestimation in upper troposphere by WRF-Chem has been reported earlier with a slightly different model setup (different convective parameterization) .

Comparison with MOPITT CO profiles
10 Figure 6 shows the monthly average CO profiles from simulation Std and the CO retrievals obtained from MOPITT over Chennai. For consistency with the comparison with CARIBIC observations (Sect. 4.1.1), which are collected only during nighttime, we restrict the comparison of WRF-Chem and MOPITT to nighttime data, though we do not find large diel variability in free tropospheric CO in our simulations. The averaging kernel and the a pri- 15 ori profiles of MOPITT data have been applied on the monthly average CO profile from standard WRF-Chem simulation, denoted as Std(AK). In contrast to the comparison with the in situ vertical profiles from CARIBIC, the WRF-Chem simulated CO shows very good agreement with the satellite data in the lower troposphere during June. The mean bias value between WRF-Chem and MOPITT is found 20 to be 1.5 ± 0.8 nmol mol −1 in the LT during June as compared to the WRF-Chem and CARIBIC data comparison (−12.6 ± 4.4 nmol mol −1 ). Interestingly, in comparison to the satellite data, WRF-Chem is found to overestimate CO in the LT by 21.4 ± 2.8, 37.8 ± 5.0, and 26.9 ± 4.0 nmol mol −1 during July, August and September respectively. Middle tropospheric CO is also significantly overestimated by WRF-Chem as compared to MOPITT 25 during July-September. This could be partially associated with the unscreened-out cloud contamination in the satellite retrievals during the summer monsoon season. The a priori CO data from the global chemistry transport model could be another potential source of the discrepancy (Asatar and Nair, 2010).
WRF-Chem profiles, after applying the satellite operator become very similar to the satellite a priori, especially in the lower and middle troposphere. During this period, averaging kernels in the lower troposphere are found be smaller (less than 0.1) as compared to the 5 values reported for example during spring (Kar et al., 2008). This indicates relatively lower sensitivity of MOPITT for lower tropospheric CO over this region during the summer monsoon. The different results regarding the WRF-Chem evaluation against the in situ measurements and satellite data clearly highlight the need of more in situ measurements of vertical profiles for validation of chemistry-transport models as well as the satellite retrievals over 10 this region, particularly during the monsoon, when the sky is obscured by clouds. Such studies would be invaluable for addressing the discrepancies due to limited overpassing time for MOPITT, retrieval errors due to sensor degradation, not updated CO a priori, cloudcontamination, systematic errors as well as errors in model simulations.

15
In order to understand if the observed discrepancies between WRF-Chem and CARIBIC observations are associated with emissions and processes at the surface in India, we analyze the variations in surface ozone and CO over this region. WRF-Chem (Std) simulated average distributions of surface O 3 and CO over the Indian region are shown in Fig. 7 for the four months of the summer monsoon in 2008. The distribution of O 3 as well as CO shows 20 large spatial heterogeneity across the region in all four months.
Surface ozone levels are typically lower (< 30 nmol mol −1 ) than aloft over most of the domain. The ozone levels are found to be highest over the polluted Indo-Gangetic Plain (IGP), in northeastern India, and also over the eastern coastal region (40-50 nmol mol −1 ).
Average surface ozone levels over most of the Indian region are relatively low, mostly be- 25 low 40 nmol mol −1 , which is mainly due to the inflow of marine air masses and suppressed photochemistry in cloudy/rainy conditions. The highest levels of surface ozone are simulated over the northern part of the domain, where the influences of marine air/monsoon are Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | relatively smallest. While, vertical trace gas distributions are affected by monsoon convection, both CO and O 3 are not soluble and not directly affected by precipitation scavenging. Wet scavenging of O 3 precursors and prevailing cloudy-rainy meteorological conditions, however, could suppress the ozone production, particularly near surface. CO mixing ratios vary from about 50 to 300 nmol mol −1 , except over the IGP where a high CO belt 5 (400 nmol mol −1 and more) accumulates throughout the monsoon season. Towards the end of the monsoon period in September, ozone and CO levels show most pronounced enhancements over the IGP and also a tendency of pollution buildup in the surrounding regions.
The WRF-Chem simulated spatial distributions of surface ozone and CO are found to be 10 consistent with previous studies over the Indian region mostly based on satellite observations (Fishman et al., 2003;Kar et al., 2008), simulations with a global chemistry transport model (Ojha et al., 2012) and previous study evaluating WRF-Chem simulations over Indian region . WRF-Chem simulations were found to significantly overestimate surface O 3 and underestimate CO at an urban site in the Indo-Gangetic Plain towards 15 the onset of monsoon, while the model was in better agreement during May (Michael et al., 2014).  (Naja and Lal, 2002;Renuka et al., 2014) and observations of CO are used from the coastal site Cape Rama (73.8 • E 15.1 • N) (Yoon and Pozzer, 2014). O 3 and CO model results from the Std simulation are found to be within the 1σ standard deviations of the measurements at Gadanki and Cape Rama.
The significant underestimation of CO by WRF-Chem in the free troposphere (Sect. 4.1.1) 25 as compared to CARIBIC measurements is not evident at the surface. It is suggested that the discrepancies between WRF-Chem and CARIBIC observations are likely not caused directly by surface emissions and chemistry and may be associated with the influence of large-scale air mass transports. We further investigate this by conducting a sensitivity simu-Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | lation with 50 % higher CO emissions (Sect. 4.2) over the Indian region. The possible role of transport is investigated by backward air trajectory analysis and conducting a sensitivity run with 25 % higher influx of CO from the domain boundary based on trajectories (Sect. 4.3).

Sensitivity to regional emissions
A sensitivity simulation 1.5×_EM has been conducted by enhancing the CO emissions 5 over the entire south Asian domain (Fig. 1) by 50 %, keeping all other inputs fixed as for the Standard WRF-Chem simulations (Std, Table 2). Previous studies (e.g. Randel et al., 2010;Fadnavis et al., 2013Fadnavis et al., , 2015 have shown that monsoonal convection plays a key role in uplifting the boundary layer emissions / pollution into the Upper Troposphere and Lower Stratosphere (ULTS) altitudes. To investigate this effect, we compare the monthly average 10 horizontal distribution of CO from Std and 1.5×_EM simulations for upper tropospheric altitudes (average for 116 − 211 hPa) (Fig. 9). The spatial distribution of CO in the upper troposphere shows highest levels in the northern and central Indian region in both of the simulations. The effects of the monsoonal circulation are clearly visible through convectively uplifted CO from regional emissions, in partic- 15 ular from the Indo-Gangetic Plain (IGP) towards the west. The sensitivity simulation shows significant influence on the upper tropospheric CO distribution and increases the westward export of pollution. For example, over the north-central Indian region the CO mixing ratios are found to be higher by about 20 % in 1.5×_EM simulation as compared to Std simulation. The comparison of monthly average CO over Chennai between the standard simulation 20 and 1.5×_EM is shown in Fig. 10. The percentage enhancement in the CO mixing ratios due to the increased emissions is also shown. The maximum impact (33 %) of the increased anthropogenic emissions on CO mixing ratios is observed near the surface. The direct impact of emission enhancement is found to be significantly lower (5 % and less) from 850 hPa and above, where WRF-Chem was found to most strongly underestimate the CO levels. 25 Hence a significant increase (50 %) in the regional anthropogenic emissions over India led to only minor enhancements in the model CO levels as compared to the observed underestimation in the lower free troposphere. Furthermore, the WRF-Chem simulated surface Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | CO is in good agreement with ground-based observations over this region. Therefore, it is concluded that the observed underestimation of CO by WRF-Chem in the free troposphere is not primarily associated with local and regional anthropogenic emissions. The next possibility of transport of CO into the domain as controlled by the chemical boundary conditions in WRF-Chem is investigated in the next subsection.

Influence of transport
We investigate the role of transport over Chennai utilizing the 10 day backward trajectories simulated using the HYSPLIT model in conjunction with a sensitivity simulation with the MOZART/GEOS5 boundary condition. Air mass trajectories color coded according to the starting altitude over Chennai for all the CARIBIC observation days are shown in Fig. 11. 10 Synoptic wind patterns appear to be very different in the lower troposphere (2-4 km) compared to higher altitudes (8-12 km). Lower tropospheric air over Chennai has been dominantly influenced by westerly air masses, while the upper tropospheric air masses primarily originated from the east during June-August. The wind patterns change significantly towards the end of the monsoon period (September), when the trajectories are influenced by 15 different continental regions of South Asia.
To investigate the transport/influences from local/regional pollution, we calculated the residence time of the air masses and mean pressure along the trajectory over the southern Indian region (74.9 to 81.7 • E and 9.9 to 17.1 • N) for all the backward air trajectories at 2 and 4 km altitude above Chennai (Fig. 12). Residence time is derived by counting number 20 of hours in the air trajectory within the specified south Indian region and converting it into days. For all days the residence time in South India was about a day, except for 15 July 2008 when the residence time was more than 3 days.

Strong pollution event on 15 July 2008
We begin by examining a pollution event observed on 15 July 2008 over Chennai to in- 25 vestigate its origin. Ozone and carbon monoxide levels were observed to be very high Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | during the month of July but are substantially underestimated by WRF-Chem ( Fig. 2 and  Fig. 3). Since there were only two observational profiles during July and both on the same day (15 July 2008), this observation is suggested to be more representative of a pollution event rather than the monthly average conditions over this region. During this event, O 3 (146.4 ± 12.8 nmol mol −1 ) and CO (136.4 ± 12.2 nmol mol −1 ) mixing ratios are found to be 5 very high in the lower troposphere (∼ 805 hPa), indicating that these concentrations are associated with the transport of polluted air with ample time for photochemical ozone build up, while significant influence of transport of ozone rich air from the stratosphere is unlikely.
It is found that the residence time of this air mass is more than 3 days over southern India during this event, much longer than during CARIBIC flight times in other months. More-10 over, the air masses are found to be influenced by boundary layer pollution as indicated by significantly higher mean pressure along the trajectory (915 ± 43 hPa). To investigate the underprediction of the event in the model, we analyzed the wind fields over Chennai from radiosonde measurements. The model is found to generally reproduce the variations in wind speed over Chennai at different altitude levels (e.g. in the range of 4-10 m/s at 980 hPa) 15 (Supplementary material- Figure 3). However, the model does not capture the occurences of low-wind speed (1-3 m/s) and overestimates systematically the wind speed during the July period. Therefore the air parcels could not possibly collect enough pollutants from the boundary layer leading to the underprediction of the strong pollution event in model. Additionally, no indication of underestimation of emissions is found as the model performance 20 did not improve in reproducing the event when emissions were increased by 50%.

Long-range transport
Long-range transport of pollution in regional models is controlled by the chemical boundary conditions, generally provided from a global model. Previous studies investigated the impact uncertainties (Andersson et al., 2015) in long-range transport in regional model simulations. In WRF-Chem simulations in this study the long-range transport is controlled by the time varying chemical boundary conditions from a global model MOZART/GEOS5 simulations. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | We assess the contribution of long-range transport of CO in the lower troposphere over Chennai by conducting a sensitivity simulation with increased CO at the domain boundary. Backward air trajectories suggest that CO is significantly underestimated in the lower troposphere in westerly air masses (Figs. 3 and 11). Therefore, we increase the CO mixing ratios by 25% in the MOZART/GEOS5 data, over a region (7.5 • N < lat < 16.5 • N) on the 5 western boundary as shown in Fig. 13, chosen suitably based on the backward trajectories (Fig. 11). Figure 14 shows a comparison of average CO profiles from CARIBIC measurements, WRF Chem standard simulations (Std) and the sensitivity simulation with increased CO at the western boundary (1.25×_BDY). In contrast to the sensitivity run with increased 10 emissions (1.5×_EM), here we find significant improvement in the WRF-Chem simulated CO in the free troposphere. For example during June, WRF-Chem simulated CO mixing ratios from 1.25×_BDY simulation (95.8 ± 4.3 nmol mol −1 ) are comparable to the observations (96.6 ± 9.1 nmol mol −1 ) at ∼ 800 hPa. The improvements are also significant in other months in the lower free troposphere. In contrast to the June-August period, the air masses 15 over Chennai show influences of higher emissions on the free troposphere. This could be associated with the transport from the continental Indian region as shown by backward trajectories (Fig. 11). The enhancements due to higher CO in the boundary conditions are significantly less during September as compared to June-August. We suggest that since air masses over Chennai during September are more influenced by the regional emissions, 20 the influence of uncertainty in boundary conditions is not evident here. Further it is noted that such dominance of regional impacts on CO vertical distributions during September is captured better by WRF-Chem, as compared to the global model simulation (Supplement Fig. 1).
This study suggests that anticyclonic advection plays a very important role which could 25 transport polluted air masses from outside the region (domain) during the summer monsoon. This complements conventional thinking that convected regional emissions dominate the tropospheric composition during the monsoon season and points to a potentially significant external source of pollution to the monsoon anticyclone. We show that this transport is generally very fast i.e. the residence time of air masses is 1-2 days over southern India, except during the strong pollution event (Sect. 4.3.1). This rapid transport could advect COrich air masses from more strongly polluted upwind regions. As indicated by the backward air trajectories and a sensitivity run, CO-rich air masses could originate in central Africa and the Persian Gulf region. During the summer monsoon, CO mixing ratios have been 5 found to be highest over central Africa associated with biomass burning emissions (Torres et al., 2010;Inness et al., 2013, and references therein). A recent study utilizing the trajectory-mapping technique and aircraft observations (Osman et al., 2015) also indicated elevated CO mixing ratios over the western boundary of our model domain. We suggest that improvements in the global fire emissions input to the global models and data assimi-10 lation would be helpful in better constraining the effects of long-range transport during the monsoon. Regional emissions from continental India are shown to significantly influence the free troposphere over Southern India towards the end of the monsoon (September).

Conclusions
In this paper we integrated the aircraft-borne measurements of O 3 and CO vertical profiles distributions. This feature has been observed to be most pronounced during July and has been qualitatively captured by WRF-Chem.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | The major limitation of the model is found to be underestimation (12.6-22.8 nmol mol −1 ) of CO in the lower free troposphere during June to August. Model simulated CO is in very good agreement with CARIBIC measurements during September. The model biases in lower tropospheric O 3 are found to vary from an overestimation by about 4.3 nmol mol −1 (June) to an underestimation by 7.8 nmol mol −1 (August). Additional simulations using a 5 different emission inventory (INTEX-B) showed similar results. WRF-Chem simulated CO is also compared with satellite (MOPITT) retrievals. Interestingly, WRF-Chem is found to overestimate CO compared to MOPITT data during July-September, while, WRF-Chem and MOPITT CO are in very good agreement during June. The contrasting evaluation results of WRF-Chem with in situ measurements and satellite retrievals points towards a need 10 of more measurements to validate the satellite data and evaluate model results over this region.
WRF-Chem simulations are also compared with the ground-based measurements of ozone and CO at Gadanki (79.2 • E, 13.5 • N) and Cape Rama (73.8 • E,15.1 • N). It is shown that the model simulated O 3 as well as CO at the surface is within the observed variabilities 15 (1σ). This indicates that the discrepancy between the WRF-Chem simulated and CARIBIC measured CO is likely not directly associated with the regional surface emissions. This is corroborated by a sensitivity simulation with 50 % higher CO emissions over India which leads to about 33 % enhancement at the surface but small influence (5 %) above 850 hPa. Neverthless, the increase in regional emissions of CO was found to influence the upper 20 tropospheric distribution over north and central Indian region, increasing westward export by about 20%.
Analysis of backward airmass trajectories and wind speed data from model and radiosonde observations over Chennai suggested that a strong pollution event observed during July (O 3 : 146.4 ± 12.8 nmol mol −1 , CO: 136.4 ± 12.2 nmol mol −1 ) was associated with 25 stagnation of regional photochemically-processed air mass over southern India.
We find that the lower free troposphere over this region is strongly influenced by air masses from the west during the summer monsoon. A sensitivity simulation with 25 % higher CO mixing ratios over a region at the western boundary of the domain, chosen Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | suitably based on back air trajectories, shows significant improvement in the computed CO levels. We suggest that long-range transport of CO over southern India, originated in Africa, is underestimated in model simulations during the summer monsoon and may have a significant impact on the regional CO budget. We recommend improvement of the global fire emissions and data assimilation in the global models to better constrain long-range trans-5 port in WRF-Chem. The effects of regional emissions and synoptic-scale transport over south Asia are better captured by the WRF-Chem. Therefore, improved boundary conditions data combined with regional model will be suitable for chemical budget studies. Our study highlights that in situ measurements limited only to the ground are insufficient to understand the transport of trace gases, and that aircraft-borne measurements of ozone pre- 10 cursors are essential to improve the model simulations and the understanding of regional tropospheric chemistry.
The Supplement related to this article is available online at doi:10.5194/acpd-0-1-2015-supplement.

Simulation Name Description
(1) Std WRF-Chem simulations driven by MOZART4/GEOS5 boundary conditions. No factor on boundary conditions or emissions. This simulation is used as the standard WRF-Chem run for this study.      (Naja and Lal, 2002) and Cape Rama (73.8 • E, 15.1 • N).
Open blue symbols for Gadanki show observations from another study (Renuka et al., 2014). Comparison with Std_INTEX simulation is indicated in black Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |