Journal of Geophysical Research: Atmospheres

Surface ozone and precursor gases at Gadanki (13.5°N, 79.2°E), a tropical rural site in India



[1] Ozone, nitrogen oxides (NO′x) (this includes NO, NO2 and some of their compounds converted by molybdenum converter; see section 2.2 for details), CO, and CH4 have been measured during the period 1993–1996 at Gadanki (13.5°N, 79.2°E), a rural site in the tropical Indian region. Observations show daytime photochemical ozone production that is initiated by the photooxidation of the precursor gases, with maximum noontime annual average O3 mixing ratios of only 34 ± 13 ppbv. Ozone levels were highest during the winter/spring period and lowest during summer, which differs from observations at sites in other regions. The rate of increase in O3 is greatest around 0900 hours local time, whereas the magnitude of the maximum rate of decrease during the evening is considerably smaller. This feature distinguishes the urban and rural sites since the magnitudes of the rates during morning and evening are more similar at urban sites. At Gadanki, annual averages of oxides of nitrogen (NO′x), CO, and CH4 are observed to be 2.1 ± 1.8 ppbv, 237 ± 64 ppbv, and 1.70 ± 0.11 ppmv, respectively. However, a poor correlation between ozone and NO′x was found, indicating that NO′x levels at this site are not controlled by fresh combustion or emissions and could be due to transport from the nearby major cities. Ozone production efficiency and estimated photochemical ozone production were also found to be lower at this site.

1. Introduction

[2] Ozone is a greenhouse gas, and it is the precursor of highly reactive hydroxyl radical (OH), that largely controls the chemical composition of the troposphere. Higher concentrations of ozone in the boundary layer have deleterious effects on biological life and vegetation [e.g., Chameides et al., 1994]. Tropospheric ozone concentrations and growth rates exhibit large spatial and temporal variabilities [e.g., Volz and Kley, 1988; Naja and Lal, 1996; Logan et al., 1999, and references therein]. Therefore, there is an uncertainty in the estimation of its contribution to the radiative forcing and tropospheric chemistry.

[3] Ozone in the lower troposphere is produced mainly by photochemistry involving pollutants that are released from various industrial and other anthropogenic activities [e.g., Fishman and Crutzen, 1978]. NOx (NO and NO2) and hydrogen oxide radicals (OH and peroxy) act as catalysts in this process. NO plays a critical role, such that ozone production may occur even in a rural region if the NO abundance is higher than a critical limit of about 10 pptv, a value that depends on ozone levels [e.g., Lin et al., 1988]. The tropospheric ozone problem was once thought to be restricted to urban areas, but now it is recognized that ozone concentrations in rural areas can rival those measured in urban areas. Study of ozone chemistry is important at rural sites because ozone precursors may get transported there from the near–by urban or industrial areas.

[4] Several studies have examined the impact of the Asian pollutants on other parts of the world [e.g., Jacob et al., 1999, and references therein]. Recently, measurements from a ship have identified transport of pollutants from the Indian subcontinent to the Indian Ocean [Lal et al., 1998]. Brasseur et al. [1998] showed that the change in tropical ozone is highly sensitive to climate forcing in the tropics. The production of OH is highest in the tropical region due to intense solar radiation and large water vapor content (80% of the global budget). This feature makes the tropical region photochemically most active [e.g., Crutzen, 1995]. In spite of the importance of the tropical troposphere, there have been no systematic simultaneous measurements of surface ozone and its precursor gases over the Indian region (except those at Ahmedabad, an urban site) until recent years [Naja and Lal, 1996; Lal et al., 2000]. Limited measurements of surface ozone are available only for the Indian region [Shende et al., 1992]. Since ozone shows large spatial and temporal variability, even on a regional scale, measurements are needed at different locations. The rate of industrial growth in India is quite large, and the number of automobiles there is increasing rapidly [Central Pollution Control Board, 1996]. In light of these important features and limited data over this region, a project was initiated under the Indian Space Research Organisation's Geosphere Biosphere Program (ISRO-GBP) for measurements of ozone and its precursor gases over the Indian region. Results from the measurements of ozone and its precursor gases made at Gadanki, a rural site in India, are presented here.

2. Observation Site

[5] The observation site was the National MST (Mesosphere Stratosphere Troposphere) Radar Facility in Gadanki (13.5°N, 79.2°E, 375 m above sea level), a rural area of Chittoor district (Andhra Pradesh state) in the southern part of India (Figure 1). The population of Gadanki is about 200–300, and it is about 30 km away from the small town of Tirupati (population about 0.2–0.3 million). There are hills in the northern and southern sides of the observation site within 1 to 10 km distance. The average height of the hills is about 750 m, with a maximum height of about 1300 m. There is no major industry in the entire Chittoor district except for few small-scale units at Tirupati. A major road passes near the observation site, with a usage of a few hundred heavy vehicles each day. The observation site is about 150 and 200 km from the two nearby major cities, Chennai (Madras) to the southeast and Banglore to the southwest, respectively.

Figure 1.

Map of the observation region along with other sites in India (inset).

2.1. General Meteorology

[6] The Southern Indian region has two monsoon seasons associated with the movement of the Inter-Tropical Convergence Zone (ITCZ) [Asnani, 1993]. The southwest (SW) monsoon occurs from June to September when the ITCZ moves northward over India, reaching up to about 30°N during July. The winter, or northeast (NE) monsoon, occurs from October to November when the ITCZ moves southward (approaching the equator). The monsoon of India is connected with that of the whole Southeast Asia. Ramage [1971] showed that the monsoon rains start over the southeast Bay of Bengal, south Myanmar and the Indo-China region by mid-May. The monsoon reaches the southern tip of the west coast of India in the last week of May, and it covers most of India by July. The winds over India are southwesterly to westerly on the southern side of the ITCZ and easterly to northeasterly on the northern side of the ITCZ [Asnani, 1993].

2.2. Measurement Techniques

[7] Measurements of ozone and nitrogen oxides are obtained routinely at 15-min intervals with analyzers that are based on the well-known techniques of UV absorption (Environment S. A., France; Model O341M) and chemiluminescence (Environment S. A., France; Model AC31M), respectively. Further details of the analyzers used in this study are given elsewhere [Naja and Lal, 1996; Naja, 1997; Lal et al., 1998, 2000]; hence only a brief description is provided here. The ozone analyzer is based on the measurement of absorption of UV radiation at 253.7 nm. The analyzer has a built-in ozone generator for auto calibration check and a procedure for zero setting. These are performed for 15 min each on daily basis for an initial period of 1 year and then once every 3–4 weeks afterward. The minimum detection limit of the analyzer is about 1 ppbv and its response time is about 10 s. The absolute accuracy of these UV absorption based analyzers is reported to be about 5% [Kleinman et al., 1994].

[8] Measurements of oxides of nitrogen are based on the chemiluminescence of NO2 produced in the reaction of NO with ozone. NO2 is measured by thermal conversion (i.e., reduction) to NO using a built-in molybdenum converter heated to 320°C. Although the conversion is about 100% efficient [Winer et al., 1974; Finlayson-Pitts and Pitts, 1986], it is known that the molybdenum converter also reduces other oxides of nitrogen such as PAN, HNO3, etc. [e.g., Winer et al., 1974]. Thus the actual concentrations of NO2, and therefore NOx, may be lower than those measured. However, it has been shown that NOx remains the major fraction of NOy (NOx + products of NOx oxidation) in rural environments, comprising usually about 40–70%, and sometimes as high as 100% of NOy [Parrish et al., 1993]. Nevertheless, the NOx measurements presented here should be considered as upper limits to their true values, and so they are represented here as NO′x. The minimum detection limit of the analyzer is reported by the manufacturer to be about 350 pptv (response time 60 s) with noise of about 170 pptv. However, we are considering only the data above 500 pptv with large signal-to-noise ratios for greater confidence. Different O3 and NO′x analyzers have been intercompared at an urban site by running these analyzers side-by-side and using a common inlet system. The correlation coefficient of O3 over the range 5–90 ppbv was found to be 0.99, and that for NO′x was 0.98 up to abundances of 75 ppbv [Naja, 1997].

[9] Air samples were collected regularly at least once per week (during selected months) in pre-evacuated glass bottles at a pressure of about 2 bar using a metal bellows pump for analyses of CH4 and CO. These air samples subsequently were analyzed using a gas chromatograph (Shimadzu, GC-mini3, Japan) equipped with a Flame Ionization Detector (FID) together with a methanizer (heated Ni catalyst) at the Physical Research Laboratory (PRL), Ahmedabad [Naja, 1997; Lal et al., 1998, 2000]. The analyses of CH4 and CO have uncertainties of about 3 and 12%, respectively. Calibrations for CH4 and CO were performed with standards from NIST, USA (SRM 1658a; 1.19 ± 0.01 ppmv) and Linde, UK (1000 ± 7.5ppbv), respectively. Each day during the measurement period these reference air samples were analyzed alternately with ambient samples.

3. Observations

[10] Continuous measurements of ozone were initiated in November 1993, and those of NO′x were initiated in January 1994. Air samples were collected in glass bottles beginning in January 1994 for the analyses of CO and CH4. Table 1 summarizes the data for all the measurements of ozone and NO′x made at Gadanki. Ozone levels in excess of 80 ppbv were observed rarely during the observation period (∼0.11% of the time), and ozone never exceeded 100 ppbv. NO′x mixing ratios never exceeded 20 ppbv during the measurement period.

Table 1. Monthly Statistical Data of O3 and NO′x at Gadanki During 1993–1996 and 1994–1995, Respectivelya
  • a

    All values are in ppbv except count, which indicates the total number of measurements of 15-min averaged data. Min and Max indicate the minimum and maximum values from 15-min averaged data.


3.1. Diurnal Variations in Ozone

[11] Figure 2 shows the monthly average diurnal variations of ozone observed at Gadanki during 1993–1996. All times are in the Indian Standard Time (IST), which is ahead of the GMT by 5.5 hours. Mixing ratios of ozone start increasing gradually after sunrise, attaining maximum values during near local noon. However, daytime maximum ozone mixing ratios exhibit changes in pattern at different times of year. Ozone can be produced during the day by photooxidation of precursor gases [e.g., Fishman and Crutzen, 1978; Lin et al., 1988; Crutzen et al., 1999, and references therein]. Later it will be shown that levels of CO, NO′x and CH4 are sufficient for this process. Ozone variability is influenced also by boundary layer processes and meteorology. The boundary layer height rises gradually after sunrise, reaching a maximum height of 1500–2000 m during midday due to convective heating. At this time, air from lower altitudes mixes with ozone-rich air from higher altitudes; it has already been shown that the contribution of boundary layer processes is significant to ozone variations at the surface [Naja and Lal, 1997].

Figure 2.

Diurnal variations of average ozone mixing ratios in different months for the 1993–1996 period observed at Gadanki. Vertical bars are 1σ variations in monthly averages.

[12] Ozone levels were observed to be lower during nighttime when there is no O3 production from photooxidation of precursors. In addition, the titration of O3 by NO in the shallow boundary layer and loss due to surface deposition continues during the nighttime. Daytime production of ozone is observed throughout the year except during cloudy and rainy days. Major rainfall occurs during the SW monsoon in June, July, and August (Figure 3a), a time when surface temperatures also decrease (Figure 3b). Therefore, the lack of sufficient solar radiation during cloudy days and influx of cleaner marine air during this monsoon period significantly reduce daytime photochemical ozone production. During the monsoon months the diurnal amplitude (peak-to-peak) of ozone is of the order 8 ± 2 ppbv (amounting to only about 40% of the average ozone levels), whereas it is 35 ± 8 ppbv during spring (amounting to about 122% of the average ozone levels). Day-to-day variability also follows a similar pattern. The maximum diurnal amplitude is observed during March when there is little rainfall. Daytime ozone maximum is observed to be sharper during the spring months, while it is broader during the rest of the year (Figure 2). The daytime maximum is observed to occur during the hours of 1400 to 1600 in March, becoming gradually broader in later months, while in November, December, and January the ozone maximum is observed during the hours of 1000 to 1600. The higher and sharper noontime peak particularly in March appears to be due to higher mixing ratios of NO′x (shown later), which lead to greater buildup of ozone during the day and also to faster titration later in the afternoon.

Figure 3.

(a) Seasonal variations in daily total rainfall, (b) temperatures (maximum and minimum of the day), and (c) daily average wind speed, observed at Gadanki during 1995. There was abnormally more rainfall for a few days in May 1995.

3.1.1. Comparison of Diurnal Variations

[13] The diurnal variations in ozone observed at Gadanki show an asymmetric diurnal cycle. Mixing ratios of ozone decrease more slowly during the evening than they increase during the morning. This type of diurnal variation in ozone, with mixing ratios remaining high until the late evening, has also been observed at the rural sites Trigg [Meagher et al., 1987] and Georgia [Kleinman et al., 1994]. This has been attributed to slower titration of ozone by NO during late evening time at these rural sites. Generally, urban sites show almost similar morning production rates and evening loss rates (i.e., symmetric variation). Figure 4a shows a comparison of diurnal variations of ozone at Gadanki with those observed at other sites in India. Measurements made at the outskirts of Pune (18.5°N) [Shende et al., 1992] show a very slow decrease in ozone mixing ratios during evening hours, similar to the observations at Gadanki. However, measurements made at Ahmedabad (23°N) [Lal et al., 2000] and Delhi (28.7°N) [Shende et al., 1992], both urban sites, exhibit symmetric diurnal variations in ozone. It is important to mention that we are only discussing the diurnal patterns in ozone. Hence, differences in absolute ozone abundances at the four sites that could be due to differences in their locations and the periods in which these measurements were made (1980s and 1990s; see Figure 4 caption) are not discussed.

Figure 4.

(a) A comparison of diurnal variations in ozone at Gadanki (13.5°N, 375 m asl) with other sites in India (Ahmedabad, 23°N, 49 m asl; Pune, 18.5°N, 559 m asl and Delhi, 28.7°N, 220 m asl). Measurements are for the period of 1983–1985 at Pune and Delhi [Shende et al., 1992] and for the period of 1991–1995 at Ahmedabad [Lal et al., 2000]. (b) Observed rates of change of ozone mixing ratios at Gadanki, Ahmedabad, Pune, and Delhi.

3.1.2. Rate of Change of O3

[14] The average rates of change of ozone during the hours 1700 to 1900 at Gadanki and Pune are estimated to be −2.6 ppbv hr−1 for both the sites, which are lower in magnitude (43–46%) than their respective production rates during the hours of 0800 to 1100 (see Table 2). Rates of change of ozone during the hours of 1700 to 1900 at urban sites (Ahmedabad and Delhi) are higher and are almost similar (8–18% in magnitude) to their respective rates during the hours of 0800 to 1100 (Table 2). This feature of ozone variations distinguishes the urban and rural sites. Since only ozone measurements were made at Pune and Delhi, levels of ozone precursor gases are not known at these two sites. NO′x levels are about ten times higher at Ahmedabad than at Gadanki.

Table 2. Observed Rate of Changes of Ozone at Four Sites in India
SitesRate of Change at 0800–1100 Hours, ppbv hr−1Rate of Change at 1700–1900 Hours, ppbv hr−1

[15] Figure 4b shows diurnal variations in the rates of change of ozone mixing ratios, [d(O3)/dt], at Gadanki, Ahmedabad, Pune and Delhi. The rates of change [d(O3)/dt] are as high as 6–7 ppbv hr−1 around the 0900 hour, except at Delhi, where rate of change is about 4.5 ppbv hr−1 at around the 1000 hour. Changes in the boundary layer height, as well as in the incoming solar radiation, are quite rapid during morning time, and therefore the maximum change in ozone is observed during this time. The [d(O3)/dt] is near zero at 1500 hour, after which it becomes negative. Ahmedabad and New Delhi (both urban sites) show sharp changes in [d(O3)/dt] during 1800–1900 hours which are not observed at Gadanki and Pune. This is consistent with faster ozone titration at urban sites, as mentioned earlier. At midnight, [d(O3)/dt] is found to be almost steady. During this time, ozone chemistry becomes less intensive and vertical mixing of air is significantly reduced by formation of a stable nocturnal boundary layer. Surface deposition likely becomes the main process for ozone loss during this period.

3.2. Diurnal Variations in NO′x

[16] Monthly average diurnal variations of NO′x for the years 1994 to 1995 are shown in Figure 5. NO′x mixing ratios are generally largest during the morning and late evening hours, although during the monsoon months the patterns are not very regular. Variations in NO′x are caused by variations in boundary layer mixing processes, chemistry, anthropogenic emissions, and local surface wind patterns. The emitted pollutants get trapped at the lower heights in the boundary layer during evening and continue to remain so until early morning due to formation of a nocturnal stable layer. However, pollutants are diluted during midday due to increased height of the boundary layer and extensive mixing. Local sources of pollutants at Gadanki are also minor. Agriculture is the major activity in this region. At this site the annual average mixing ratio of NO′x is found to be about 2 ppbv only, which is small compared to urban sites, where mixing ratios can be as high as 20–50 ppbv.

Figure 5.

Diurnal variations of average NO′x mixing ratios in different months for the 1994–1995 period observed at Gadanki. Vertical bars are 1σ variations in monthly averages. There were no observations during September.

3.3. Seasonal Variations in Ozone

[17] Figure 6 shows the seasonal variations in ozone at Gadanki for daily averages at the hours of 1200–1400, 0100–0300, and for monthly averages. The average mixing ratios of ozone for 1200–1400 hours and monthly averages show a maximum value during winter/spring and a minimum value in summer-monsoon. After the monsoon period the mixing ratios of ozone increase during late autumn and maximum value is in March (Table 1) with maximum variability (1σ standard deviation) also in the same month. However, solar radiation at this latitude is not greatest in March. There are some spikes in average ozone mixing ratios during 1200–1400 hours in April and May, but these do not contribute significantly to their respective monthly averages. Variability in ozone mixing ratios is lower during the monsoon period than during other seasons. Interestingly, the decrease in ozone variability (1σ standard deviation) from March to August is larger (about 68%) than the decrease in the corresponding monthly average values (about 44%). This makes it clear that variability in March is higher compared to the reference of change in monthly mean ozone mixing ratio. This suggests a possible role of some unknown process during March, which is discussed later. Average mixing ratios of ozone during 0100–0300 hours do not show the variations that are observed during 1200–1400 hours. Because the boundary layer height rises to 1500–2000 m during midday, ozone can be mixed up to this height due to convective heating. Therefore, average ozone mixing ratios during 1200–1400 hours could be representative of the upper boundary layer and free lower tropospheric ozone levels.

Figure 6.

Seasonal variation of ozone for daily average of 1200–1400 hours, 0100–0300 hours, and monthly average at Gadanki. Vertical bars in monthly average are 1σ variations.

[18] Vertical ozone measurements made during INDOEX-IFP (Indian Ocean Experiment-Intensive Field Program) over the Arabian Sea and the Indian Ocean in February–March 1999 show many instances of higher ozone in the middle troposphere than near the surface. These results also show that stratosphere-troposphere exchange (STE) plays an important role in the tropical tropospheric ozone budget [Zachariasse et al., 2000]. Lamarque et al. [1999] suggested that the STE could contribute to about 10–20% in the tropical lower troposphere. We feel that ozone levels at Gadanki in spring, particularly during 1200–1400 hours, might be influenced by transport from above. Possible contribution of emissions in this regard is discussed later.

[19] Figure 7 shows a comparison of seasonal variations in ozone observed at Delhi, Ahmedabad, Pune, and Gadanki. In general, all four sites exhibit low ozone mixing ratios during the monsoon season. However, this effect extends into October at Gadanki. This is likely due to the fact that Gadanki also experiences the winter monsoon. After the monsoon, ozone concentrations increase sharply at both Gadanki and Ahmedabad. It has been shown that this increase in ozone levels during late autumn and early winter at Ahmedabad is due to the increase in abundances of precursor gases that lead to photochemical ozone production during these seasons [Lal et al., 2000]. The highest ozone levels occur during February and March at Gadanki, which is not the case for other sites in India. Most of the continental sites at midlatitudes in the United States, Canada, and Europe exhibit maximum ozone during summer or spring seasons, which is believed to be due to intense photochemical ozone production during these seasons [e.g., Logan, 1985]. Interestingly, measurements made at Delhi show maximum ozone mixing ratios during spring/summer [Shende et al., 1992]. These differences are related to the availability of solar radiation and high concentrations of precursor gases, which are caused by local emissions or transport from the regional sources.

Figure 7.

A comparison of seasonal variations in ozone observed at Gadanki (13.5°N, 375 m asl) with those at other sites in India (Delhi, 28.7°N, 220 m asl; Ahmedabad, 23°N, 49 m asl; Pune, 18.5°N, 559 m asl).

3.4. Seasonal Variations in NO′x

[20] Seasonal variations of NO′x averages for the hours 1200–1400, 0100–0300, and monthly means are shown in Figure 8. The monthly mean NO′x is greatest during March, similar to ozone. However, NO′x mixing ratios during November and December are not observed to increase as does monthly mean ozone. Nighttime NO′x levels are often higher than levels during midday (as discussed earlier while discussing the diurnal variations in NO′x). Unfortunately, measurements of NO′x were less regular than those of ozone during the monsoon period.

Figure 8.

Seasonal variations of NO′x for daily average of 1200–1400 hours, 0100–0300 hours, and monthly average at Gadanki. Vertical bars in monthly average are 1σ variations.

3.4.1. Ozone Production Estimate

[21] Figure 9 shows a scatterplot of daily mean O3 with NO′x for four seasons. There is a weak positive relationship (correlation coefficient of 0.3) with slope of 3.3 for the entire data set, but relationship is different for different seasons. It is positive for winter and spring seasons and negative for summer and autumn seasons. The slope of this correlation is related to photochemical production or production efficiency of O3 (ppbv) per unit NO′x (ppbv). Table 3 provides the details of regression parameters for the four seasons. Estimates made by Kelly et al. [1984] and Glavas [1999] using a similar NO2 conversion technique (heated molybdenum) for NOx measurements, show ozone production efficiencies of 6–7 O3 (ppbv) per unit NO′x (ppbv). Ozone production efficiency at Gadanki is lower compared to these studies for midlatitudes. Olszyna et al. [1994] showed that lower O3 production efficiency and weaker correlation are indicative of partially processed air entering the sampling region and rule out fresh emissions, which is the case for Gadanki. Higher correlation is expected only for those air masses in which ozone production has just occurred [Kleinman et al., 1994].

Figure 9.

Average ozone mixing ratios for different seasons as a function of average NO′x for the measurements made during 1994–1995.

Table 3. Regression Parameters for O3 and NO′x Correlation for Measurements Made During 1994–1995 at Gadanki
SeasonsMean ± 1σ O3, ppbvMean ± 1σ NO′x, ppbvIntercept,a ppbvSlopeaCorrelation Coeff.Number of Days
  • a

    Standard errors are shown in parentheses.

Winter (DJF)29.2 ± 6.31.8 ± 0.526.0(2.3)1.8(1.2)0.1595
Spring (MAM)30.5 ± 10.22.5 ± 0.818.4(2.9)4.9(1.1)0.40106
Summer (JJA)31.5 ± 7.72.0 ± 0.633.5(9.9)−1.0(4.7)−0.089
Autumn (SON)28.6 ± 5.31.7 ± 0.539.1(5.1)−6.4(3.0)−0.6011
Total29.9 ± 8.42.1 ± 0.822.9(1.6)3.3(0.7)0.30221

[22] We feel that it is more appropriate to consider the spring data set because the correlation coefficient is relatively higher for spring and more measurements are available during this season. Considering the ozone production efficiency (4.9) for spring and the NOx emissions for India given by van Aardenne et al. [1999], we estimate ozone production of about 14–22 Tg (O3)/yr for the Indian region. However, for the entire data set (ozone production efficiency of 3.3) it is about 11 Tg (O3)/yr only. These estimates are lower than that of 100 Tg (O3)/yr estimated for the East Asia using a similar approach [Jaffe et al., 1996, and references therein]. However, our estimates are a first-order approximation only since the present data set is very small, it is for one station only and the correlation coefficient between ozone and NO′x is poor.

3.4.2. CO and CH4 Measurements and Emissions

[23] The annual average mixing ratios of CO and CH4 are 237 ± 64 ppbv and 1.70 ± 0.11 ppmv respectively at Gadanki (Figure 10). These species also show lower mixing ratios in the rainy season. The seasonal amplitude (peak-to-peak) in CO is observed to be about 240 ppbv, which is considerably larger (about a factor of 2–6) than that observed at a cleaner site, Mauna Loa, Hawaii.

Figure 10.

Measurements of CH4 and CO in different months during 1994–1996. The air samples were collected and subsequently analyzed by gas chromatograph for CH4 and CO. In the box plot, inside thin and thick lines are median and mean, respectively. Box boundaries are 25th–75th percentiles, whiskers are 10th–90th percentiles, and tilt squares represent all the data out side of 10th–90th percentiles.

[24] Biomass burning is the major source of CO. Most biomass burning (about 85%) takes place in the tropical countries but the contribution from tropical Asia is only about 16% of the total [Hao and Liu, 1994]. India is second to China for various emissions and energy consumption in the Asian region [Akimoto and Narita, 1994; Hao and Liu, 1994; van Aardenne et al., 1999]. Biomass burning occurs mainly from January to May, peaking in February, March, and April in the Indian region [Galanter et al., 2000]. Maximum emissions of CO and NOx over India from biomass burning are estimated to occur during this period. However, the contribution of biomass burning to NOx production is less over India and China than over Africa, such that the biomass burning contribution to the ozone budget should also be smaller over India. Its maximum contribution is only 10–20% during March, April, and May whereas it is up to 50% in other tropical regions during their biomass burning seasons [Galanter et al., 2000].

[25] Biomass burned in the southern Indian states (Andhra Pradesh, Karnataka, Kerala, and Tamil Nadu; commonly called southern India) is very small (1.4 Tg/month) compared to the eastern Indian states (6–7 Tg/month) and the western Indian states (4.6 Tg/month) [Hao and Liu, 1994]. Southern India contributes only about 20% to the budgets of SO2, NOx, and CO2 on the national level [Akimoto and Narita, 1994]. Furthermore, the southern region has less vehicular pollution. The three major cities in the southern region (Banglore, Chennai, and Hyderabad) contribute much less NOx (71 tonnes/day), CO (465 tonnes/day), and hydrocarbons (185 tonnes/day), compared to Delhi, which alone contributes 126 tonnes/day, 651 tonnes/day, and 250 tonnes/day, respectively (1tonne = 1000 kg) [Central Pollution Control Board, 1996]. Therefore, the southern Indian region does not seem to be a large source of emissions as compared to the rest of India.

4. Discussions and Summary

[26] Measurements at Gadanki show production of ozone during daytime due to photooxidation of the precursor gases. However, asymmetry in diurnal variation is observed with higher ozone levels persisting until the late evening hours, which is not observed at urban sites. The observed rate of decrease in ozone during evening time at Gadanki is found to be smaller than its morning time production rate.

[27] The seasonal variation in ozone at Gadanki is different from that observed at other observation sites in different regions. Relatively higher levels of ozone observed during February and March at Gadanki could be due to prevailing biomass burning during this period and possible transport of precursor gases from the major nearby cities like Chennai and Banglore to this region. Downward transport from upper altitudes may also contribute to these higher ozone levels, but there is no direct evidence for this. The general climatology over this region shows that the surface wind is northerly or northeasterly from November to March period [Ramage, 1971; Asnani, 1993]. Therefore, winter and spring period may experience relatively polluted air from the north and northeast region. The change in wind pattern occurs during May and continues until September/October. During this period the wind becomes westerly or southwesterly, carrying the pristine air from the Arabian Sea and the Indian Ocean. In general, surface wind speeds (Figure 3c) follow a trend similar to that observed in ozone and other gases. Other meteorological parameters are not available for this site to discuss their role in detail in the present study.

[28] Emissions are lower over the southern Indian region compared to the rest of India. The effect of biomass burning, which mainly occurs from January to May, is not reflected in the measurements of CO. However, NO′x, which is produced mainly by fossil-fuel combustion, shows higher levels during spring and particularly in March. Therefore we suggest that higher NO′x emissions in the presence of other hydrocarbons could lead to a higher production of ozone during March. During February, when ozone levels are higher (but not those of CO and NO′x), the observations suggest that other processes, such as downward transport of ozone, are important.

[29] Average ozone levels at Gadanki are lower compared to many global sites such as northern United States, Mexico, Europe, and also China, where ozone levels exceeding 80 ppbv for many days have been commonly observed [e.g. Aneja et al., 1999; Raga and Raga, 2000; Luo et al., 2000]. There has not been any case of ozone levels exceeding 80 ppbv even for 3 hours continuously at Gadanki. In fact, there were only 10 hours total in March 1994, the month of highest ozone levels, when ozone levels exceeded 80 ppbv. In addition, abundances of 100 ppbv were never attained. We find a poor correlation of about 0.3 only between ozone and NO′x. Ozone production efficiency estimated using ozone-NO′x correlation is lower at Gadanki than at other sites in midlatitudes. Such a weak correlation in O3 and NO′x, and lower O3 production efficiencies are consistent with the absence of local emissions and highlight the role of long-range transport process at this site. However, there is a need for more and extensive measurements of ozone and precursor gases for a better understanding of the processes controlling ozone variability over this region.


[30] We are grateful to B. H. Subbaraya for his encouragement and fruitful discussions. We thank P. B. Rao and A. R. Jain, past and present directors, respectively, of the National MST Radar Facility (NMRF), Gadanki, for providing the observational site. Our thanks are owed to Y. Bhavani Kumar of NMRF for regular maintenance of the analyzers. We greatly appreciate the help provided by K. S. Modh, S. Venkataramani, T. K. Sunil, and S. Desai in making these measurements and data analysis. We are grateful to R. Ramesh for critically going through the manuscript. We also thank the two reviewers for their many constructive suggestions, which have improved the quality of this paper. This work is carried out under the ISRO-Geosphere Biosphere Programme.