Assessment of Spatiotemporal Variability and Rising Trends of Surface Ozone Over India

Ravi Kumar Kunchala (  rkkunchala@cas.iitd.ac.in ) Indian Institute of Technology Delhi Bhupendra Bahadur Singh IITM: Indian Institute of Tropical Meteorology Karumuri Rama Krishna KAUST: King Abdullah University of Science and Technology Raju Attada IISER Mohali: Indian Institute of Science Education and Research Mohali Vivek Seelanki IIT Delhi: Indian Institute of Technology Delhi Kondapalli Niranjan Kumar National Centre for Medium Range Weather Forecasting


Introduction
Surface ozone has become a widely studied constituent in recent times due to its emergence as a short-lived (~ few days) secondary pollutant and its damaging impacts on human health and crop production (e.g. Mills et al. 2007;Avnery et al. 2011;Ghude et al. 2014;Lelieveld et al. 2015). It plays a pivotal role in tropospheric chemistry through controlling the oxidation processes as the source of hydroxyl radicals (OH) (Levy, 1971;Logan, 1985) and through its reaction with many organic compounds (Atkinson, 1990). Ozone in the troposphere is a secondary air pollutant product of photochemical reactions involving carbon monoxide (CO) and volatile organic compounds (VOCs) in the presence of nitrogen oxides (NOx = NO + NO2) termed as ozone precursors (Chameides and Walker, 1973;Fishman et al. 1979;Crutzen, 1995). These photochemical reactions are dependent on solar radiation and usually, the rate of reaction is highest during the summer months. As compared to surface ozone, the ozone concentrations are higher in the upper tropospheric levels where it is affected more by the large-scale circulations and exchanges between stratosphere-troposphere (Brunamonti et al. 2018; Kumar et al. 2021). The surface ozone has a relatively shorter life span and exhibits signi cant spatiotemporal heterogeneity (Ojha et al. 2012;Nair et al. 2018). Recent studies have reported an increasing surface ozone trend over different parts of the world including Asia, where a rise in concentrations poses a stiff challenge to more than 5 billion people (Monks et al. 2015;Singh et al. 2021b). The fast-paced developing economies such as China and India have experienced enhanced anthropogenic emissions which have led to increased ozone concentrations in recent decades (Cooper et al. 2014;Sun et al. 2016;Wang et al. 2017). This rise has been noted through various precursors e.g., biomass burning, fossil fuel combustion, and lightning, etc.
Understanding the spatiotemporal variability of surface ozone over India is important as it ranks amongst the toxic air pollutants and any scienti c literature discussing these aspects remains elusive. A few studies have investigated the surface ozone variability over different parts of India and the surrounding Indian Ocean region (e.g., Nair et al. 2011;Peshin et al. 2017;Tyagi et al. 2016;Tiwari et al. 2015;Girach et al. 2017;Anshika et al. 2021). Most of these studies have been focused on particular observational locations and therefore a comprehensive understanding of spatial distribution and variability is yet to be explored. The long-term analysis of surface ozone over India has been limited by the availability of reliable and consistent records. Although satellite-based measurements provide tropospheric surface ozone concentrations, they are often of coarse temporal and spatial resolution (Ojha et al. 2012). Model-based analysis has reported signi cant biases along with limitations in capturing small and regional scale variability (see Sharma et al. 2017 and references therein). A recent study by Hakim et al. (2019) analyzed the observations and outputs from model simulations to understand the annual cycle of surface ozone over India. However, the analysis was constrained by the limited number of observational locations and time. Therefore, it requires spatially well sampled and reliable estimates to correctly examine surface ozone variability and impacts. Some studies have thus highlighted the need for an increased network of ground-based stations (Schultz et al. 2017).
In this regard, ozone records based on station observations coordinated by the Central Pollution Control Board (CPCB) India, provide a good source for validating remotely sensed and reanalysis datasets and study the subregional scale variability over different parts of India. Another important aspect is that India has complex geography/topography and is subjected to strong seasonal variations in circulation patterns.
Signi cant seasonal changes in the circulation patterns are seen during different seasons e.g., from premonsoon to monsoon, monsoon to post-monsoon followed by post-monsoon to winter months annually.
Driven primarily by the incoming solar radiation, strong changes in circulation and convection leads to the annual cycle dominated by the monsoon type of climate (Singh et al. 2021a). Therefore, the region is subjected to complex interactions between the regional meteorology and anthropogenic emissions which can combinedly contribute to the spatial and temporal variability of surface ozone.
With a rise in population and anthropogenic activities, it is important to understand the surface ozone variability over the region. Though studies have documented its variability over isolated locations, any comprehensive and consistent study is lacking thus far. It is imperative to analyze the distribution and seasonal behavior of ozone over India and understand its variability at longer time scales. Keeping these facts in mind, the present study primarily examines surface ozone variability across India based on the observed and reanalyzed datasets. We rst compare the reanalysis data with in situ high resolution station records over different geographical subregions (north, east, west and south). A comprehensive validation exercise is carried to infer the reliability of Copernicus Atmosphere Monitoring Service (CAMS) data and quanti cations are presented and discussed. Through the time series analysis, we study the seasonality of surface ozone over different locations across India. Further, we use the CAMS data to assess the spatiotemporal variability of surface ozone across India. Additionally, we analyze the associated meteorology and brie y discuss the linkages of meteorological variables with seasonal variations in surface ozone. Basically, the present study aims towards bringing out the statistics of surface ozone over the Indian region as seen through the lens of seasonal variability, its relationship with meteorological variables of interest, and identifying the dominant modes of variability. The organization of the paper is as follows: Sect. 2 discusses the data and methodology; Sect. 3 discusses the results of analyses followed by the summary and conclusions in Sect. 4.

Data And Methodology
In the present study, we have utilized ground-based observations and reanalysis datasets over India to address the surface ozone variability. The details of ground-based observations and the reanalysis data products is described in the following subsections.

Central Pollution Control Board (CPCB) station observations
CPCB measures ambient air quality through a network of stations across India under the National Air Quality Monitoring Program. Under this program, the surface ozone is regularly monitored through automatic monitoring stations out of which we have considered about 20 ground-based stations over different parts of India for the period 2018-2019. The CPCB data is available in public domain (https://app.cpcbccr.com/ccr/#/caaqm-dashboard-all/caaqm-landing/caaqm-comparison-data). We choose the stations objectively so that the data is well distributed in space across India and also there is continuity in time. This is done to understand the seasonal cycle and validate the reanalysis-based dataset. It also helps in analyzing small-scale regional in uences. Therefore, we use data only from those stations with at least > 80% records available during the analysis period. The site speci cs including latitude, longitude, and altitude (height above the sea level) of stations used in the analysis are provided in Table 1. Table 1 List of ground-based stations, their location (latitude, longitude and height above the mean sea level), and the metrics of comparison used for the validation of CAMSRA data. CAMSRA data is shown to be useful in capturing the climatology, trends, and also in evaluating chemistry-based models (Inness et al. 2019). For the present study, we utilize the CAMSRA derived surface data of ozone for the period 2003-2019, archived at 0.75° x 0.75° horizontal resolution. Before using CAMSRA data for the analysis, it is validated against the CPCB ground station-based observations. Additionally, we use the reanalyzed meteorological elds from the fth generation European Centre for Medium-Range Weather Forecasts (ERA5) (Hersbach et al. 2020) to understand the association of surface ozone with meteorology.

Validation of CAMSRA
Analysis of globally averaged CAMSRA surface ozone shows that it compares well against the ground-based surface observations available from the World Meteorological Organization (WMO) coordinated Global Atmosphere Watch (GAW) programme (see Inness et al. 2019 and references therein). The study also reports a good agreement across the European region. In general, the surface ozone elds from CAMSRA and observations match well within 10% for most years with biases being negative during the rst half of the year and positive during the second half. The study robustly concluded that CAMSRA is more consistent than the previous reanalysis and has smaller biases compared with the independent ozone observations.
However, we notice that there are only a few stations available over India under the WMO GAW program with differing periods of data availability. Therefore, we rst validate the daily averaged surface ozone values for the period of the recent two years (2018 and 2019) from the latest generation CAMSRA against CPCB ground-based measurements and quantify the robustness of agreement. The stations used for this purpose are shown in Fig. 1 and also listed in Table 1. These ground-based (20) Table 1) are arranged in such a way that the latitude decreases as we move downwards. We notice that the surface ozone concentrations vary in the range ~ 2-80 ppb for both CAMSRA and CPCB observations for most of the stations. However, a relatively larger spread in these values over the northern and eastern parts of India e.g., over Ambala, Delhi, Faridabad, Gurugram, Varanasi, and Gaya. We compute the correlation coe cient (CC) and note that CAMSRA based surface ozone estimates are well correlated with the stations located in the southern and the western parts (correlation coe cient > 0.6) e.g. ppb for Hyderabad and Amaravati respectively. It is interesting to note that the range of surface ozone over the coastal station Thiruvananthapuram is slightly lesser (~ 3-51 ppb for observations and ~ 11-60 ppb for CAMSRA) as compared to those in the higher latitudes, but the MAE (15.26 ppb) and RMSE (14.28 ppb) are still higher. This is indicative of signi cant differences between the day-to-day variations between the two datasets.
Next, we analyze the time series of surface ozone obtained from CAMSRA and compare it with ground-based observations as shown in Fig. 3. For this, we have considered four representative ground-based stations taken from each of the four boxes shown in Fig. 1. These stations are located over northern, eastern, western, and southern parts of India and help represent distinct variability over polluted and coastal locations. ppb. The station-based records show similar seasonality, but with lesser amplitudes compared to the CAMSRA. As compared to the other three stations, the Thiruvananthapuram station shows a shifted peak in time which is observed during the late winter and early summer months (~ 50 ppb) while the minimum is seen during the later part of the monsoon season (~ 20 ppb). A similar surface ozone cycle was also reported by Nair et al. (2018) wherein the authors attribute the minimum to moisture-laden strong westerly/south-westerly winds which bring cleaner air mass, and a decrease in available solar radiation under cloudy monsoon conditions. The study ascribes an increase in surface ozone to the transport of continental pollutants (mostly precursors of ozone) by easterly/north-easterly winds during the winter to early summer months. Analysis of surface ozone time series indicates that the CAMSRA has the capability for reproducing the seasonality irrespective of the location. It reasonably captures the temporal evolution of the highest and lowest concentrations of surface ozone when compared with the ground-based observations. Also, we notice frequent occurrences of higher concentrations as we move ahead in time, which probably indicates an increasing trend in the surface ozone.
Though we have discussed these metrics based upon stations located across India, it is also important to note that surface ozone is signi cantly affected by local factors. Therefore, even though two locations may be close in space, the variability may differ signi cantly. This analysis quanti es that the southern stations are in better agreement with the CAMSRA while the stations over northern and eastern subregions show relatively poorer agreement. Such differences may be caused by increased emissions of precursors over the region (Verma et al. 2017). It is to be kept in mind that these values might fall closer on the seasonal and annual time scales if we use long-term observations. Nevertheless, CAMSRA reasonably captures the minima, maxima, and seasonality of surface ozone over most of the locations. Ojha et al. (2019) carried out a similar analysis for an observatory located in Dehradun and showed the comparison between groundbased observations and CAMSRA. While comparing the noontime surface ozone they found that it varies between 11.6-83.7 ppb and 46.4-111.6 ppb for observations and CAMSRA respectively. They obtained a CC of 0.86 and a mean bias (RMSE) of about 29.5 ppb (31 ppb) and concluded that CAMSRA was able to capture the day-to-day variations in noontime ozone. We however realize that such results may differ regionally which would be explored in the later sections. Nevertheless, our comprehensive analysis brings out the extent of coherency between CAMSRA and CPCB station observations across India and proves that CAMSRA can be reasonably used to examine the daily, monthly and seasonal variability in the absence of ground-based observations. The validation and time series analysis performed here equips us to utilize longterm (2003-2019) CAMSRA data to understand the spatiotemporal variability of surface ozone and its linkages with the meteorology over the region. Hereafter the analysis of surface ozone would be based on the CAMSRA while that of meteorological parameters would be based on ERA5 reanalysis.

Climatological spatial distribution of surface ozone
We use the long-term CAMSRA dataset and obtain the monthly climatology which is shown in Fig. 4. The maxima in surface ozone are usually seen over a broad region covering northern and western parts, and peninsular India during the pre-monsoon months (March-May) while minimum over these regions is observed during monsoon months (June-August) after which it starts increasing again. Increase during the May month is more prominent over the Northern Indian region especially over Indo-Gangetic Plains (IGP).
The highest surface ozone concentrations vary mostly in the range between 40-60 ppb while the lowest values range between 15-25 ppb. During the pre-monsoon season, it occasionally crosses 70 ppb over the northern stations. However, there are regions e.g., the western and southern tips where the maximum and minimum do not co-occur in time with the larger domain. Over the eastern and western arcs, we observe pronounced surface ozone concentrations during the winter months. The reason behind higher concentrations during the pre-monsoon months is linked to the availability of solar radiation and precursors which supports photochemical ozone production. This aspect would be discussed in subsequent sections (see Fig. 6). The observed increase during the winter months might be due to the stagnant atmospheric conditions (see Verma et al. 2017). We note a marked decrease in surface ozone during the monsoon months over southern and south-western parts with concentrations dropping close to 15 ppb. It is seen that with the progression of the monsoon, low values spread over most parts of India. Studies have reported that the onset of the South Asian summer monsoon in late May and early June brings westerly and southwesterly winds from the Arabian Sea to the Bay of Bengal resulting in rainy weather conditions over the Indian region (Gadgil, 2003;Yadav and Singh 2017). The resulting monsoon precipitation e ciently removes ozone precursors (see Kumar et al. 2012), wherein cloudy conditions and low temperatures hamper ozone production through photochemical reactions. We note that the surface ozone minimum over the northeastern region persists until the post-monsoon months as the region continuously experiences rains during this time. These characteristics are also noted when we analyze the seasonal mean behavior of surface ozone across India during different seasons (see Fig. S1) viz. winter (December-January-February), premonsoon (March-April-May), monsoon (June-July-August), and post-monsoon (September-October-November). To understand the spatial variability of surface ozone we also analyze the standard deviation (SD) climatology (Fig. 5). This helps in identifying the regions which experience more variability in surface ozone concentrations. High variability is seen over northern (and eastern coastal regions) parts of India especially over IGP regions during May and June with values above 5 ppb. During May, higher variability is also seen over the southern peninsula. The lowest variability with values of about 2-3 ppb across India is seen during the core monsoon months (July-August). The variability gradually increases during the postmonsoon months and continues for the winter months. Clearly, the ozone variability is in uenced by the monsoon with a minimum during the rainy months and higher values at other two ends viz. the pre-and post-monsoon months.
We further analyzed the area-averaged climatological concentrations over four geographical regions namely Northern India (NI), Eastern India (EI), Western India (WI), and Southern India (SI) shown as boxes in Fig. 1 (see Fig. S2). Climatologically, we note that the highest concentrations of ozone over NI (EI) are observed in May with values around ~ 46 ppb (~ 49 ppb). Interestingly we note that the minimum is observed during January (August) for NI (EI), which shows a distinct ozone seasonality over these regions. The climatological mean is smallest over WI where the highest (lowest) concentrations occur during April (August), with a value close to ~ 36 ppb (~ 24 ppb). Among the four regions, the southern region stands out in terms of seasonality.
Here the maximum (~ 48 ppb) surface ozone is observed during the winter month, however, the minimum (~ 26 ppb) is noted during the monsoon season (August). Overall, the analysis indicates the distinct seasonality over different subregions of India during the study period.

Linkages with regional meteorology
In this subsection, we discuss the meteorological parameters of interest to surface ozone production and variability. We have evaluated the climatology of surface temperature and solar radiation (SR) and the results are shown in Fig. 6. An obvious one-to-one correspondence is seen between solar radiation and temperature. In both parameters we note a clear seasonality with maximum values during pre-monsoon months where temperature (SR) values vary between 34-38°C (800-1000 W/m2) across India. These premonsoon conditions are conducive for ozone formation which can be seen in the spatial distribution of surface ozone (Fig. 3) wherein higher solar radiation (and temperature) helps in the chemical production of ozone through the photochemical reactions. The photochemical production is further enhanced by an increase in ozone precursors (e.g. NOx) (Jacob and Winner, 2009;Doherty et al. 2013;Pusede et al. 2015), and natural emissions such as biogenic NMVOCs and soil NOx emissions (Lu et al. 2018) during the premonsoon season.
With the arrival of monsoon, a gradual decrease in temperature (SR) from June is seen and the lowest values of less than 20°C (~ 550 W/m2) start appearing during the winter months (from November) over northern parts. The temperature (SR) further decreases over the northern regions of India with values ranging between 8-18°C (300-500 W/m2) during the peak winter months. The north-south gradients in temperature and SR become more prominent as we move from post-monsoon to winter months. During the monsoon months, cloudy conditions and associated precipitation lead to a reduction in solar radiation and surface temperature, thereby reducing photochemical ozone production. The winter-time high concentration of surface ozone over the northern, eastern, and IGP regions is linked primarily to an increase in CO levels. The CO increase during this season is caused by a combination of factors which include direct emissions (Hakim et al. 2019) and contributions from secondary sources such as the oxidation of Volatile Organic Compounds (Grant et al. 2010).
Next, we analyze the seasonal cycle of surface wind circulation (shown by stream ow) and boundary layer height (BLH) evolution across India. These factors are important as they give an idea about the potential transport and mixing of ozone in the troposphere. It can be seen that pre-monsoon months are associated with increased BLH and continental north-westerly (NW) winds (as seen in the stream ow patterns) over the western and northern parts of India (Fig. 7). Higher BLH co-occurs with higher surface ozone concentrations while NW stream ow brings dry air which further raises the temperature over the region making the conditions conducive for surface ozone production (Yadav et al. 2016). The south-westerly (SW) ow during the monsoon season brings moist oceanic air from the Arabian sea and cloudy conditions over the region which limits the photochemical production and reduces the BLH. The appearance of surface ozone minimum over the southern tip and west coast is consistent with the ow patterns in the early phase of the monsoon. A further drop in surface ozone over the western parts and peninsular India is seen during the core monsoon months of July and August. During this phase, the Bay of Bengal branch of monsoon circulation brings moist air over the eastern and north-eastern parts which result in cloudy conditions and reduces the surface ozone production. The post-monsoon to winter increase in surface ozone seems to co-occur with low BLH and the advent of north-easterly (NE) ow. It potentially results in the transport of precursors over the region and helps in ozone transport and formation. In general, low BLH favors the accumulation of pollutants while high BLH supports the vertical mixing of the precursors which can help in the production assisted by strong insolation and surface temperatures (Yadav et al. 2016). Low BLH in winter months may result in accumulation and stagnation of pollutants near the surface, which also prevents the mixing of ozone poor surface air with the ozone rich air present at higher altitudes. Our analysis indicates that monthly (or seasonal) evolution of surface ozone is in uenced by the availability of SR and higher surface temperature which aids in photochemical production, along with enhanced or suppressed convective mixing and transport by winds.

Dominant modes of surface ozone variability over India
In order to further examine the spatial and temporal variability of surface ozone, we have performed the Principal Component Analysis (PCA) and the results have been shown in Fig. 8-11. The usage of PCA is advantageous as it reduces dimensionality without losing important information in large datasets. This will be used to explain the relationship between dominant modes of spatial variability and their temporal evolution. Figure 8 shows the variance explained by different independent modes obtained from the PCA over the period 2003-2019. For the four different seasons considered, the rst mode shows a high percentage variance explained ranging between 30-50% while the variance explained by the second mode ranges between 10-20%. Hence the rst two modes together explain large variance (40-70%) within the data. Therefore, we primarily discuss the spatial and temporal patterns of the rst two modes only. For instance, Fig. 9 shows the spatial patterns of the rst mode of PCA (PCA-1). It indicates that during the pre-monsoon (MAM) conditions, much of the variance is explained from the south and central Indian region while in other seasons much of the variance can be seen from the central and eastern Indian region.
Speci cally, during the monsoon, the core monsoon region shows a high variance of surface ozone, which might be linked to cloudy and rainy conditions. Further, the spatial distribution of PCA-1 is also closely related to the climatological distribution of surface ozone shown in Fig. 4. The spatial distribution corresponding to the second mode of PCA (PCA-2) is shown in Fig. 10. It is interesting to note the premonsoon spatial distribution of PCA-2, where the IGP region shows high variance. During the monsoon period (JJA) and post-monsoon season (SON), we see a dipole type of pattern. For the monsoon season, this pattern is seen over the west coast and the central Indian region, while for the post-monsoon season it is seen over south India and northwest India respectively. These spatial patterns of the rst two modes explain much of the spatial variability seen in the climatological surface ozone distribution shown in Fig. 4. It is also interesting to understand the temporal evolution of these two modes which is shown in Fig. 11. In Fig. 11(a), one can see the time series of PCA-1 mode which indicates an increasing trend irrespective of the season.
This implies that over much of the south and central Indian region the surface ozone has increased during 2003-2019. The time evolution of the second mode indicates some contrasting trends in different seasons. For instance, during the pre-monsoon (post-monsoon) season where much of the ozone variability is associated with the IGP (southern India), we do not see any signi cant trend instead we observe a signi cant interannual variability. While it is interesting to note the dipole spatial pattern of PCA-2 associated with the monsoon season (Fig. 10), it shows a signi cant decreasing trend (Fig. 11b). Interestingly the second mode during JJA which shows a decreasing trend and is associated with a dipole spatial pattern, resembles the regions of active (organized) and break (weak) monsoon spells during the season (Rajeevan et al. 2010;Singh et al. 2021a). Overall, much of the variance is primarily explained by the PCA-1, which indicates an increasing trend of surface ozone over the Indian region.

Summary And Conclusion
Surface ozone is one of the major toxic secondary pollutants and is important to understand its distribution and variability. In the present study, we have examined these aspects of surface ozone over India using CPCB station observations and CAMSRA datasets during 2003-2019. Before using CAMSRA, we have validated CAMSRA against 20 ground observations across different subregions of India. We have performed the PCA and investigated the spatiotemporal variability of surface ozone and linkages with the meteorological parameters. Following are the main highlights and conclusions of this study: 1. The validation analysis indicates that the CAMSRA surface ozone is in good agreement with the stationbased observations. It is noted that the correlation coe cients for different subregions at 95% level of signi cance are: ~ 0.6 for southern and the western parts, and ~ 0.4 for northern and eastern parts of India. The range, MAE, and RMSE have also been computed for all the stations used in the analysis. We nd that the correlation is signi cant, and there is a fair agreement in the range. However, we note that CAMSRA overestimates the surface ozone particularly the lower part of the spectrum. Overall, the validation analysis through CC, MAE, and RMSE brings out the extent of coherency between CAMSRA and CPCB data, and results indicate that it can be reasonably used for the analysis of long-term surface ozone variability and trends over India.
2. Time series analysis of daily averaged surface ozone from CAMSRA and observations shows strong seasonality in surface ozone concentrations. Despite overestimations, we note that CAMSRA compares well against observations at each station irrespective of the location. Over the polluted northern station e.g., Delhi, surface ozone usually peaks in May with maximum concentrations frequently reaching beyond 60 ppb while the minimum is seen during August when the concentrations drop below 20 ppb every year. Surface ozone variability over another polluted eastern station e.g., Varanasi is similar to that of Delhi (except the winter minimum), where it peaks in May (~ 60 ppb) and drops to a minimum during August (~ 20 ppb). The time series over the western region e.g., Udaipur shows the maximum (minimum) surface ozone during April (September) with values ranging between 35 ppb to 40 ppb (15 ppb to 20 ppb). The southern subregion e.g., Thiruvananthapuram shows a clear seasonality with values varying between (maximum of) 50 ppb and (minimum of) 20 ppb. 3. Spatiotemporal distribution of surface ozone shows maximum over a broad region covering northern and western parts, and peninsular India during pre-monsoon months (March-May), while minimum concentrations are observed during the monsoon months (June-August). An increase during the premonsoon month of May is more prominent over the Northern Indian region, especially over the IGP. The highest surface ozone concentrations vary mostly in the range between 40-60 ppb while the lowest values range between 15-25 ppb. During the pre-monsoon season, it crosses 70 ppb occasionally over the northern parts. The observed increase in surface ozone concentrations during the pre-monsoon months is due to increased SR and temperature which favors the photochemical ozone formation in the presence of precursors. 4. Analysis of meteorological parameters e.g., temperature, SR, winds (stream ow), and BLH show coherent seasonality where enhanced levels of surface ozone during the pre-monsoon months are linked with high surface temperature (SR) varying between 34-38°C (800-1000 W/m2). These conditions favor ozone formation through photochemical reactions (which is con rmed from the ozone distribution maps). During the monsoon months, cloudy conditions and precipitation lead to a reduction in solar radiation and surface temperature thereby reducing photochemical ozone production. 5. Analysis of PCA shows that the rst two modes together explain large variance in the surface ozone with variance explained ranging between 30-50% and 10-20% respectively. The rst mode indicates that during the pre-monsoon (MAM) conditions, much of the variance is explained by the south and central Indian region while in other seasons it is mostly contributed by the central and eastern Indian region. The time series of PCA-1 mode indicates an increasing trend irrespective of the seasons. The second mode during JJA shows a decreasing trend which is associated with a dipole spatial pattern resembling the regions of active and break monsoon spells during the season. The time evolution of the rst and second modes shows a contrasting trend with signi cant interannual variability. . The second mode during the pre-monsoon (post-monsoon) is majorly associated with the IGP (southern) region. The temporal evolution of the second mode indicates large interannual variations indicative of strong meteorological controls on the surface ozone variations over these regions.

Declarations
Ethical approval and consent to participate: Authors consciously assure that the manuscript is an original work and follows the ethical standards of research.

Figure 2
Scatter analysis highlighting the correlation (value mentioned in the brackets) between ground-based observations and CAMSRA surface ozone over different locations (as shown in Fig. 1) across India.

Figure 4
Climatological spatial distribution of monthly surface ozone based on CAMSRA during the period 2003-2019. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Same as Figure 4, but for standard deviation. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Same as Figure 4 but for solar radiation (shaded) and temperature (contour). Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Same as Figure 4 but for Planetary Boundary Layer Height (shaded) and stream ow (contour). Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.  Spatial pattern of PCA mode 1 (PCA-1) for different seasons: (a) Pre-monsoon, (b) Monsoon, (c) Postmonsoon, and (d) Winter. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 10
Similar to Figure 9 but for PCA mode 2 (PCA-2). Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.