Features of upper troposphere and lower stratosphere aerosols observed by lidar over Gadanki, a tropical Indian station

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113,, doi: /2007jd009411, 2008 Features of upper troposphere and lower stratosphere aerosols observed by lidar over Gadanki, a tropical Indian station Padmavati Kulkarni, 1 S. Ramachandran, 2 Y. Bhavani Kumar, 1 D. Narayana Rao, 1 and M. Krishnaiah 3 Received 22 September 2007; revised 9 April 2008; accepted 23 May 2008; published 10 September [1] Upper troposphere (UT) and lower stratosphere (LS) aerosol characteristics are studied over a tropical station Gadanki (13.5 N, 79.2 E), using 532-nm Nd:YAG lidar during Scattering ratios (SR) and aerosol extinction are found to exhibit seasonal and interannual variations in UT (10 15 km) and LS (18 30 km). SR is about in the 10- to 30-km altitude region. Aerosol extinction is about km 1. SR in UT in 2001 and 2004 during winter is lower than that of summer, whereas LS winter profiles are found to have higher SR values. SR values apparently experience a shift in altitude corresponding to the seasonal change in tropopause indicating a relation between the two. UT integrated extinction is about 2.5 times higher than LS extinction. The correlation between UT and LS monthly mean aerosol extinction was weak and negative with a coefficient of 0.4. UT and LS aerosol extinctions over Gadanki are found to exhibit an increasing trend during The percentage contribution of integrated aerosol extinction in the 10- to 30-km region to aerosol optical depth (AOD) is about 12%. Correlation coefficient between monthly mean AOD and the 10- to 30-km integrated extinction is about 0.6. The increasing trends in UT and LS aerosols seem to support the finding that emissions from subtropical and tropical Asia may have already started to influence the amount of sulfur containing gases reaching UT and LS. This becomes relevant in the climate change context as it has been shown that increasing aerosol abundances may be impacting the monsoon. Citation: Kulkarni, P., S. Ramachandran, Y. Bhavani Kumar, D. Narayana Rao, and M. Krishnaiah (2008), Features of upper troposphere and lower stratosphere aerosols observed by lidar over Gadanki, a tropical Indian station, J. Geophys. Res., 113,, doi: /2007jd Introduction [2] Optical and physical characteristics of aerosols vary as a function of altitude and residence time. The stratospheric aerosols are quite different from the tropospheric aerosols. The aerosols in the stratosphere have a longer residence time of a few months to years when compared to about a week in the lower troposphere. Tropospheric aerosols are short-lived due to gravitational settling and rainwash, produce regional and seasonal effects, whereas stratospheric aerosols are long lived and produce long-term global effects. Vertically resolved aerosol physical and optical properties are important to understand the role aerosols play in altering the radiation budget of the Earth s atmosphere. The tropical upper troposphere (UT) and the lower stratosphere (LS) is a crucial region for understanding aerosol characteristics and stratosphere troposphere exchange. 1 Department of Space, National Atmospheric Research Laboratory, Gadanki, Andhra Pradesh, India. 2 Space and Atmospheric Sciences Division, Physical Research Laboratory, Ahmedabad, India. 3 Department of Physics, S.V. University, Tirupati, India. Copyright 2008 by the American Geophysical Union /08/2007JD [3] Banded structures in stratospheric aerosol distributions were seen during both quiescent and volcanic periods [Trepte et al., 1994]. Maximum optical depths were found over the tropics and high latitudes while minima were encountered in the 15 to 45 latitude region [Trepte et al., 1994]. The observed aerosol reservoir over the tropics was attributed to troposphere stratosphere exchange, which gets aided by the enhancement in lofting associated with the easterly shear of quasi biennial oscillation (QBO). The vertical transport of mass between the stratosphere and troposphere is a crucial process in atmospheric physics and chemistry. The timescales of vertical mixing distinguish the troposphere from the stratosphere; throughout the troposphere transport of air and chemical species can occur in few hours via strong updrafts associated with cumulus formation typically in the vicinity of Intertropical Convergence Zone (ITCZ), while it may take months to years for vertical transport over a similar altitude range in the stratosphere [Seinfeld and Pandis, 1998]. The largest net upward transport into the stratosphere occurs in the tropics which consequently can affect the global middle atmosphere [Rosenlof, 1995]. [4] In addition the tropical UT is the region of high nucleation rate and newly formed particles, and can survive 1of11

2 Table 1. Number of Days of Lidar Observations Over Gadanki During , the Data of Which are Used in the Present Study Month January February March April May June July August September October November December Total until carried aloft to the stratosphere [Hamill et al., 1997]. The LS aerosol layer is strongly dependent on input from the tropical UT in the absence of major volcanic eruptions. In the absence of volcanic eruptions, both natural and manmade gaseous sulfur sources are responsible for the formation of a background stratospheric sulfate layer [Hofmann, 1993]. Aerosol concentrations decrease above the tropical tropopause due to coagulation during the upward air transport produced by stratospheric circulations [Brock et al., 1995]. These results show the importance of UT and LS aerosols in the tropics for the sustenance of aerosols in higher latitudes, thereby highlighting the need for long-term measurements over the tropics to document the increase or decrease in aerosol behavior. [5] Long-term studies on aerosols have been mostly confined to midlatitudes using lidars [Osborn et al., 1995; Jäger et al., 1995] and balloon-borne optical particle counters [Hofmann, 1993; Hervig and Deshler, 2002]; global view of aerosol characteristics was possible mainly in the LS with satellite limb extinction instrument measurements [Trepte et al., 1994; Hitchman et al., 1994; Kent et al., 1998; Thomason and Taha, 2003]. There have been few studies in the tropics using lidars, including Mauna Loa (19.5 N) in Hawaii [Barnes and Hofmann, 2001] and Naha (26.2 N) in Japan [Uchino et al., 1995]. The evolution of Pinatubo volcanic aerosol layer and its decay was documented with a newly set up Nd:YAG lidar at Ahmedabad (23 N), India from April 1992 to May 1994 [Jayaraman et al., 1995]. However, over India so far no long-term data on the vertical profiles of aerosols during volcanically and quiescent conditions are available. [6] In this study, the high-powered lidar system installed at Gadanki (13.5 N, 79.2 E), a tropical station in south India, is used to study UT and LS aerosol profiles with good temporal and vertical resolution. The observations began in 1998 and are still in progress. The present study covers the period , during which more than 400 days of measurements were made using 532 nm ground based lidar (Table 1), which were analyzed to study UT and LS aerosol behavior. Since there has been no major volcanic eruption after June 1991 (Pinatubo) and as the residence times of volcanic aerosols is about three years, the aerosol characteristics measured beyond 2000 would correspond to volcanically quiescent conditions [Ramachandran and Jayaraman, 2003]. Aerosol characteristics during volcanically quiescent condition help examine the long-term trend that reflects the natural variability of aerosols in the UT and LS layers. In this study the seasonal and altitudinal variations in aerosol characteristics determined during volcanically quiescent conditions are examined and discussed. 2. Site and System Description [7] Gadanki is a tropical rural site located at an altitude of about 375 meters above mean sea level and about 80 km from the Bay of Bengal coast, in southern India. To understand atmospheric changes with reference to the monsoon conditions the aerosol characteristics are classified into four seasons viz., winter (December, January, and February), pre-monsoon (March, April, and May), summer (June, July, and August), and post-monsoon (September, October, and November). In summer at Gadanki, the minimum and maximum surface temperatures range from about 22 C to43 C; during summer, convective activity is more when compared to any other season in a year while in winter the temperature varies from about 13 Cto32 C. [8] The elastic backscatter lidar system was made operational in The transmitter part consists of Nd:YAG pulse laser source at the second harmonic of 532 nm with a maximum energy of about 550 mj per pulse. The laser operates at a pulse width of 7 nsec with a repetition rate of 20 Hz. The receiver consists of a portable, compact, and vertical Schmidt Cassegrain telescope with a diameter of 35 cm. The field of view (FOV) of this system is 1 mrad. A photo multiplier tube with a narrowband interference filter centered at 532 nm (FWHM of 1.13 nm) is used in front of the polarization beam splitter. The polarized beam splitter splits the beam into cross- and co-polarized components. Optical attenuators are kept in between the beam splitter and photo multiplier tubes (PMTs) for two independent channels, which are designated as P (co-polar) and S (crosspolar) containing PMTs of comparable gains. The photon counting signals are sent to an MCS-Plus channel for signal analysis. The MCS-Plus is a full size plug in card, which records the counting rate of events as a function of time. The lidar system is a monostatic biaxial system and is optically aligned to heights greater than 8 km so that lowlevel clouds and aerosols would not interfere with the observation. The dwell time of the counting system is 2 msec, which corresponds to an altitude resolution of 300 m. The backscattered returns (photon counts) summed for 250 sec which correspond to 5000 laser shots are stored in a computer hard disk through the data acquisition system. 3. Observations, Data Analysis, and Sensitivity [9] The lidar at Gadanki is operated only during clear sky conditions, thus restricting the number of days of observation in a month. Typically about 4 6 hours of observations are made during each observation day. Table 1 shows the number of observational days as function of month from 2001 to 2005, the data of which are used in the present study. There are either less or no observations during monsoon period mainly due to limitations of optical system for operating during clear sky conditions (Table 1). [10] An elastic backscatter lidar detects the total atmospheric backscatter without separating the aerosol and 2of11

3 molecular contributions to the backscatter signal. In order to separate aerosols from molecules inversion methods are adopted. Klett inversion method [Klett, 1985] has been used extensively for lidar retrieval of aerosol characteristics and the same is used in the present study. The method is executed on the 532-nm co-polarized lidar signal to obtain aerosol optical properties, such as extinction coefficient and scattering ratio. Klett inversion method assumes a constant backscatter to extinction ratio and calculates the total backscattering coefficient. The backscattered photon counts are first corrected for background counts, which is done by averaging the number of data points from the far end of the signal. The background corrected signal is then range corrected. The noise and range corrected photon counts are normalized with the model atmosphere [U.S. Standard Atmosphere Supplements, 1966] air density profile corresponding to 15 N, typically in the altitude range of about 35 km where backscattered photons due to aerosols are either insignificant or nil. [11] From the reference altitude z m (=35 km) where the backscattering contribution is mainly due to air molecules, the top to bottom integration is performed. The total backscattering coefficient b z obtained using the top to bottom inversion algorithm proposed by Klett can be written as, b z ¼ 1 b m þ 2 expðs S m Þ Z Zm Z expðs S m Þdz B a ð1þ where S and S m are logarithm of the range-corrected photon counts at any altitude z and the reference altitude z m respectively. b m corresponds to the Rayleigh backscattering coefficient at altitude z m. Rayleigh backscattering coefficients are estimated from pressure and temperature profiles for U.S. Standard Atmosphere at 15 N for 532 nm. B a (sr 1 ) is the ratio between aerosol backscattering and extinction coefficients. If B a is taken as a constant and not dependent on altitude as is done in the present study, then the above solution is similar to that obtained by Fernald [1984]. Using lidar and simultaneous balloon-borne optical particle counter data over Garmisch (47.5 N) Jäger and Hofmann [1991] obtained values of B a for 532 nm during , which includes the effects of the 1982 El Chichon volcanic eruption. However, for the background conditions using the same data excluding volcanic period data, Ramachandran and Jayaraman [2003] calculated B a values. B a values were found to be in the range of with a mean value of about The mean value of B a is used in the present study to calculate total backscattering coefficient b z. The uncertainty in the derived aerosol extinction using the above B a values was found to be about 20 25% in the 15- to 30-km region, while the uncertainty was less than 20% in the altitude region below [Ramachandran and Jayaraman, 2003]. [12] b z, total backscattering coefficient is further used to calculate the Scattering Ratio (SR). SR is a primary parameter derived from the analysis of the lidar measurements while studying aerosol and cloud properties. The lidar scattering ratio, defined as the ratio of total (aerosol plus Rayleigh) backscattering coefficient b z to Rayleigh backscattering coefficient b m is analogous to number mixing ratio of the molecules, and can be expressed as, SR ¼ b mðþþ z b a ðþ z b m ðþ z where b m (z) and b a (z) are the Rayleigh and aerosol backscattering coefficients at altitude z respectively. Aerosol extinction coefficients b ext (km 1 ) are obtained by assuming a constant ratio of aerosol extinction to backscattering called the Lidar Ratio (LR) and corresponds to 1/B a. In this work LR of 52.6 is used for calculating the aerosol extinction coefficients. b a ðþ¼ z ðsr 1Þb m ðþ z b ext ðþ¼lidarratio z b a ðþ z [13] Lidar profiles measured over Gadanki during each observable night have a range resolution of 300 m and time resolution of 250 sec. The Nd:YAG lidar at Gadanki is well calibrated and validated in many studies [e.g., Sivakumar et al., 2003; Parameswaran et al., 2004; Sunilkumar and Parameswaran, 2005]. The signal-to-noise ratio of lidar is poor up to the desired reference altitude of about 35 km especially during the presence of cirrus clouds. To get sufficient signal-to-noise ratio daily lidar measurements are time-integrated over 4 6 hours of observation time. SR and corresponding b ext are calculated for such daily averaged profiles. Two typical profiles are shown in Figure 1 as examples. Figure 1 shows vertical profiles of SR and b ext obtained on 9 May 2002 (a, b) and 22 March 2004 (c, d). [14] For an aerosol-free Rayleigh atmosphere, SR is unity and with increasing aerosol concentration or clouds SR value increases. SR is typically found to be in the range during volcanically quiescent or background conditions [Barnes and Hofmann, 2001; Jayaraman et al., 1995]. SR in May 2002 shows enhanced value of about 5 at 12 km altitude region indicating the presence of a thin cirrus cloud, while SR is about 4 in the 13- to 15-km region suggesting the presence high altitude cirrus. The corresponding b ext (Figure 1b) is about km 1 at the cirrus cloud altitudes and decreases in the LS to about km 1 at 25 km and reaches a minimum of km 1 at 33 km. On 22 March 2004 a non-cirrus day, SR was less than 1.2 and while b ext was close to molecular extinction value in the UT (Figure 1d). In the lower stratosphere b ext again decreases to about km 1 at 33 km. [15] In the analysis, SR of 1.25 is taken as a threshold for separating cirrus days from non-cirrus days. We have adopted this criterion to separate SR due to aerosols from that due to cirrus based on the threshold value. The 1.25 threshold is taken in our algorithm so that whenever and wherever cirrus clouds (SR > 1.25) are present the values are replaced by 1.25 only at cirrus altitudes without affecting the other altitudes. Gadanki being a tropical station, experiences cirrus clouds in the UT (Figure 2a) frequently during the observation period, thus hampering the aerosol vertical profile measurements. For deriving aerosol profiles, information on the cirrus cloud base and top occurrence ð2þ ð3þ ð4þ 3of11

4 Figure 1. Scattering ratios (a, c) and aerosol extinction coefficients (b, d) measured over Gadanki on cirrus (9 May 2002) and non-cirrus (22 March 2004) days. frequency is essential. In Figure 2a the frequency of occurrence of cirrus in terms of cloud base (solid) and cloud top (dotted) altitudes during are depicted at 1 km interval in the altitude region of 9 to 19 km. In case of multiple cirrus layers that appear occasionally, cirrus base and top altitudes are separated and used in the analysis. The higher frequency of about 22% cirrus top occurs in the 16- to 17-km altitude near the tropopause, the average height of which is around 17 km. The distribution of maximum cirrus base frequency is about 15% and confined to km. Above 17 km it is seen that the frequency of cirrus cloud occurrence decreases drastically to less than 5%. Cirrus cloud statistical studies conducted earlier [Sivakumar et al., 2003; Parameswaran et al., 2004] over the same site Figure 2. (a) Frequency of occurrence of cirrus over Gadanki during Cloud base (solid) and top (dotted) altitudes in the 9- to 19-km altitude region are shown. (b) Scattering ratio profiles on non-cirrus days during summer (June July August, n = 3) and winter (December January February, n = 12) 2004 over Gadanki. 4of11

5 Figure 3. A composite of scattering ratio profiles on cirrus and non-cirrus days during summer (June July August, n = 25) 2004 over Gadanki. also reported the maximum of cirrus is confined to 8 17 km and are generally closer to the tropopause. [16] In Figure 2b, SR profiles obtained on non-cirrus days during summer (JJA) and winter (DJF) 2004 are shown. The profiles clearly show that on non-cirrus days SR values are not higher than 1.25 during both summer and winter over Gadanki. The threshold value of 1.25 for SR was chosen after analyzing the 5-year ( ) non-cirrus cloud data over Gadanki. SR values are not expected to exceed 1.2 during non-cirrus days as the lower stratosphere and upper troposphere is in quiescent state. As an example and to illustrate further the capability of the lidar and the sensitivity aspect in SR, vertical profiles of SR are plotted for summer 2004 (JJA, 25 days, Table 1) in Figure 3. The tropopause height is around 17 km during summer [Satheesan and Krishna Murthy, 2005] and it is clearly seen that above the tropopause height the aerosol layer is uniform. However, below the tropopause due to the presence of cirrus clouds, a large variation is seen. In some cases cirrus layers are seen above the tropopause indicating incursion of cloud tops into the stratosphere. 4. Results and Discussion [17] Monthly mean and interannual variation in aerosol characteristics corresponding to the upper troposphere and lower stratosphere during are discussed. Aerosol extinction coefficient, scattering ratio, integrated extinction in the upper troposphere (10 15 km) and lower stratosphere (18 30 km) are investigated. Monthly mean contribution of integrated aerosol extinction in the 10- to 30-km altitude region to the aerosol optical depth is also examined during using lidar and MODIS data Upper Troposphere [18] Upper troposphere aerosols are important in understanding the global change and long-range transport mechanisms. It is difficult to study them through satellite and instrumented aircraft [Kent et al., 1998] due to high altitude clouds, while studies on upper troposphere aerosols by balloon-borne techniques are sparse. Consequently the variations in upper troposphere aerosol are much less recorded when compared to stratospheric or lower tropospheric aerosol [Kent et al., 1998]. In this context lidar has a distinct advantage over other techniques in the long-term measurements of UT aerosols. Monthly mean evolution of SR and b ext as a function of altitude are shown for 2001 and This is done because during the 5-year period ( ) only in 2001 and 2004 vertical aerosol profiles are available for all the months in a year (Table 1); the features discussed here are more or less similar in the other years. [19] The study revealed a large seasonal variability in the upper tropospheric SR and b ext. SR is higher than 1.2 during June September as these months are dominated by cirrus clouds. However, during October December, SR values are less than 1.2 with a clear increase as altitude increases from 10 to 15 km. Figures 4c and 4d contours show b ext as a function of month and altitude during 2001 and b ext is found to vary as a function of altitude during both the years; b ext varies from to km 1. b ext values are higher in the km and then decrease gradually till 15 km. b ext in UT is found to show a dip in May 2001 while during 2004 b ext and SR values are lower during January March. Kent et al. [1998] found that the variations in aerosol extinction depend on interannual variability in aerosol source strengths and meteorological factors, apart from volcanic influence and uniform seasonal changes Lower Stratosphere [20] Figure 5 contour plots show mean SR and b ext for year 2001 and 2004 respectively in the 18- to 30-km altitude region. SR and b ext showed prominent layers typically having thickness of about a km, indicating the presence of stratified layers beginning near the tropopause and extending up to 30 km during background conditions. SR is found to be about in the 18- to 30-km region. SR and b ext are comparatively high during the months of March April May 2004 in the 18- to 26-km region. b ext in LS is found to vary from a low of about km 1 to a high of km 1 with higher b ext in the 18- to 20-km region. It is interesting to note that SR values are low in km region during January March 2004, while SR is high during March May 2004 in the 18- to 26-km region. During background conditions there can be variations in aerosol extinction, as at any rate, a very slow change in aerosol size due to condensation will be masked by the larger-size changes related to the absorption and evapora- 5of11

6 Figure 4. Monthly mean aerosol scattering ratios and extinction coefficients in the upper troposphere region of km during 2001 (a, c) and 2004 (b, d). tion of water due to variations in temperature [Hamill et al., 1997] Seasonal Variation in Vertical Profiles of Aerosols [21] To understand further the seasonal variations in aerosol characteristics SR values are plotted from 10 to 30 km for winter and summer seasons for the years 2001, 2004, and 2005 in Figure 6. Horizontal bars represent ±1s and are plotted at 2 km interval for purposes of clarity and illustration. The winter and summer variation in SR is found to exhibit a relation with tropopause altitudes. We have used Vaisala radiosonde observed temperature profiles to determine cold point tropopause (CPT); CPT can be identified as the altitude at which temperature is minimum. Radiosonde observations were conducted every day at 12:00 GMT at Gadanki during (unpublished data). The mean tropopause altitude for summer (77 days of observations) and winter (51 days of observation) were found to be 17 km (temperature K) and 17.3 km (temperature K) respectively. We did not have such radiosonde measured data during and we expect that no significant change in the tropopause altitude would have occurred between 2001 and These results are quite consistent with earlier results obtained on tropopause altitudes during different seasons using Mesosphere Stratosphere Troposphere (MST) Radar. [22] Tropopause heights over Gadanki have been derived using MST Radar operating at 53 MHz. The vertical wind data were used to compute the altitude profiles of temperature following the method of [Revathy et al., 1996]. The method involves identification of Brunt Vaisala frequency (N) from the altitude profile temperature which is obtained by integrating the surface temperature as the boundary value. The method is applicable only under convectively Figure 5. Lower stratosphere scattering ratios (a, c) and extinction coefficients (b, d) measured during 2001 and 2004 over Gadanki. 6of11

7 Figure 6. Winter (December January February) and summer (June July August) scattering ratio profiles over Gadanki during (a) 2001, (b) 2004, and (c) Horizontal bars on the profiles indicate ±1s from the mean. stable condition where N 2 is positive. Satheesan and Krishna Murthy [2005] showed that the seasonal tropopause was colder in winter than in summer over Gadanki. The mean tropopause altitudes in the winters of 1999, 2000 and 2002 were 17.7, 17.4 and 17.5 km respectively; the summer tropopause altitude during 2001 was 17.2 km. The SR in UT for corresponding 2001 and 2004 during winter is lower than that of summer, while in LS winter SR are higher than summer SR in 2001 and UT and LS winter and summer SR are more or less similar in the 10- to 30-km altitude region. This may be due to the January 2005 eruption of Manam volcano in Papua New Guinea (4.10 S, E). The values of SR are also higher during 2005 at all altitudes as compared to 2001 and The ±1s variation in SR in LS during winter and summer overlap each other in 2001, while the 2004 winter and summer SR profiles are separated significantly. [23] After the tropopause altitude, (1) a general decrease in SR is seen both during summer and winter, and (2) SR values are higher during winter than in summer during 2001 and 2004, though this effect is not seen in SR values apparently experience a shift in altitude corresponding to the seasonal change in CPT indicating a relation between SR and tropopause altitude, as has been seen earlier by Rosen and Hofmann [1975]. Hamill et al. [1997] reasoned that the winter high and summer low at LS could be due to the stratosphere warming up in summer and cooling in winter. As the temperature decreases, the aerosol droplet can maintain equilibrium with the surrounding water vapor only by growing more dilute, that is by absorbing water molecules. Therefore, the particles will grow larger in winter and shrink in summer [Hamill et al., 1997], resulting in higher aerosol extinction in winter in the LS region. Another possible cause for the increase and higher aerosol extinction during winter over the tropics in LS could be due to the net upward mass flux across a pressure surface in the tropics which is twice larger during winter than in summer [Rosenlof, 1995]. In 2001 and 2004 these above features could have resulted in winter summer differences in LS aerosols over Gadanki, while in 2005 owing to the minor volcanic eruption the winter summer differences are not significant Trends in UT and LS Integrated Extinction [24] Figure 7a shows the integrated aerosol extinction in the two altitude regions of UT (10 15 km) and LS (18 30 km) as a function of month during In order to investigate the increase or decrease in aerosol extinction over the five year period trend lines are plotted using least squares fit method. The integrated extinctions are found to exhibit different seasonal variation in these two altitude regimes; UT extinction is minimum during winter (DJF) and maximum during summer (JJA), while LS extinction is higher during winter in general. The seasonal pattern in integrated extinction in UT that of winter low and summer high is more clearly seen in and 2005, while in 2004 a significant increase in UT integrated extinction is observed since April and the extinction is more or less the same throughout the year from April. The trend line shows that the integrated extinction in UT tends to increase during the study period. The percentage increase in integrated extinction in the UT is about 5%.The seasonal variations in LS are not clear; integrated extinction in LS shows more prominently a winter high, for example, 2001 and The LS extinction also shows increasing trend with an increase of about 17% during the 5-year period. [25] The observation period of was marked by highs in SR and aerosol extinction due to minor volcanic eruptions and some unknown sources. On 3 November 2002 at El Reventador (0.077 S, W) volcano erupted, which was the first in 26 years and was possibly the largest 7of11

8 Figure 7. Time evolution of (a) integrated aerosol extinction in the upper troposphere (10 15 km) and in the (b) lower stratosphere region of 18 to 30 km. Trend lines plotted using least squares fit method are shown. eruption ever seen in Ecuador [Reischmann et al., 2003; Thomason and Taha, 2003] which could have resulted in higher SR values measured during November 2002 January 2003 (Figure 5). During 2004 April and May an enhancement was noticed in SR and extinction throughout the 10- to 30-km column over Gadanki (Figures 4 and 7) which could be due to the influx of particles into the upper troposphere and lower stratosphere from an unknown source. Such a phenomenon of an increase in aerosol mass was observed in the 15- to 30-km altitude region over Ahmedabad (23 N) 30 months (January 1994) after Mt. Pinatubo volcanic eruption [Jayaraman et al., 1995]. [26] During February 2005, an enhancement is seen in SR and aerosol extinctions (Figure 7) which can be attributed to the January 2005 eruption of Manam volcano in Papua New Guinea (4.10 S, E). Note that as the frequency of occurrence of cirrus is only about 2% above 18 km (Figure 2) the increase in SR and aerosol extinction during November 2002 and February 2005 could be attributed to the minor volcanic eruptions. Aerosol extinction and SR peaked at km for 2 3 days and higher SR values have been observed throughout the column (Figures 7a and 7b). The volcanic aerosol layer produced due to Manam volcano was found to have a vertical thickness of 1 km at the altitudes of 19 km in the regions of 0 2 N and 7 9 N along about 156 E over the tropical western Pacific [Kamei et al., 2006]. [27] The increasing trend in integrated extinction during in the UT region is quite consistent with the earlier results obtained over the Indian subcontinent using SAGE II data. The upper troposphere (5 15 km) integrated extinction at 525 nm over the Indian subcontinent (5 35 N, E) from 1985 to 2000 were analyzed [Ramachandran and Jayaraman, 2003]. The upper tropospheric integrated extinctions were in the range of and a modest increase of 3% per year in the upper troposphere aerosol extinction was found [Ramachandran and Jayaraman, 2003]. Analysis of 10-year aerosol backscatter coefficient data set corresponding to at Pasadena, California (34 N) [Menzies and Tratt, 1995] showed little long-term trend. The stratospheric aerosol layer above MLO did not show a statistically significant long-term trend in the background aerosol [Barnes and Hofmann, 2001]. The increasing trend in aerosol extinction over Gadanki is in contrast to the results obtained from long-term aerosol data which showed no significant change [SPARC, 2006]; however, our trend analysis is only from a 5-year data set and could be affected by minor volcanic eruptions. [28] The increase in anthropogenic emissions could have influenced the aerosols in the upper troposphere over India [Ramachandran and Jayaraman, 2003]. The sulfur dioxide and aerosol emissions from coal were found to have increased by about 10% per year over India during The possible causes for the increase in the background stratospheric levels could be the increase in industrial emissions of sulfur dioxide [Bekki and Pyle, 1994]. Aircraft are found to generate far less aerosol than that emitted and produced at the earth s surface or by strong volcanic eruptions [IPCC, 1999]. Global changes in sulfate aerosol properties at subsonic air traffic altitudes were also found to be small over the last few decades. Long-term data suggest that aircraft operations up to the present time have not substantially changed the background aerosol mass [IPCC, 1999]. [29] The global distribution and the amount of stratospheric sulfate aerosols may change depending on the increase or decrease of precursor gases such as carbonyl sulfide (OCS) and sulfur dioxide (SO 2 ), changes in their emission patterns, and circulation changes associated with changing climate [SPARC, 2006]. SO 2 originates from both natural (volcanoes, oceans, biomass burning) and anthropogenic (fossil fuel) sources resulting in a distribution that has strong geographical signature. Anthropogenic emissions contribute approximately 70% [IPCC, 2001]. Deep convective uplift is the primary mixing mechanism for tropospheric 8of11

9 Figure 8. (a) 550-nm derived aerosol optical depth from MODIS Terra satellite over Gadanki compared with (b) the 532-nm integrated aerosol extinction in the 10- to 30-km region over Gadanki during SO 2 which is particularly efficient in the tropics during summer [SPARC, 2006]. The future trends of anthropogenic sulfur released in the tropics may affect the amount of upper tropospheric SO 2 available for upward transport in the stratospheric tropical pipe [SPARC, 2006]. Aerosols in the upper troposphere are not only made up of H 2 SO 4 H 2 O droplets, but also compose of organics, mineral dust, soot and other compounds [SPARC, 2006]. The emissions from subtropical and tropical Asia including China and India may have already started to influence the amount of sulfur containing gases reaching the stratosphere [Notholt et al., 2005]. OCS in the upper tropical troposphere may also have increased due to increases in biomass burning [Notholt et al., 2003]. The lower stratospheric aerosol layer, according to model sensitivity studies, is found to depend on inputs from the upper troposphere [SPARC, 2006]. [30] The upper troposphere lower stratosphere coupling in aerosol characteristics is examined over Gadanki using the 5-year lidar data. The time evolution of UT and LS integrated extinction (Figure 7) was utilized to obtain the monthly mean UT and LS integrated extinction. A scatter between the integrated extinctions in UT and LS yielded a weak correlation coefficient of 0.4. At Pasadena a strong coupling between UT and LS aerosols was observed and it was noted that LS was the predominant source for the UT aerosols [Menzies and Tratt, 1995]. At Pasadena the UT and LS integrated extinctions were comparable, while in Gadanki UT values are higher suggesting that UT aerosols could be a source of LS aerosols Aerosol Optical Depth and Integrated Extinction [31] The Moderate resolution Imaging Spectroradiometer (MODIS) is the primary imager on the Earth Observing Systems satellite. The Terra satellite is in a sun-synchronous near polar orbit of 705 km and views the entire surface of the earth every one to two days [Remer et al., 2005]. In this study daily mean aerosol optical depths (AOD), Level 3 Collection V005 product from Terra platform available at 1 1 latitude-longitude resolution are used. From the daily mean aerosol optical depths monthly mean aerosol optical depths at 550 nm are derived for Gadanki (13.5 N, 79.2 E) [Ramachandran, 2007]. Monthly mean AODs for Gadanki during are plotted in Figure 8a. Monthly mean AODs obtained have been validated against in situ Aerosol Robotic Network (AERONET) measurements over Kanpur (26.4 N, 80.3 E) [Ramachandran, 2007]. [32] The integrated aerosol extinction from 10 to 30 km derived from 532-nm lidar is plotted in Figure 8b for comparison. The contribution of aerosols in the 10- to 30-km altitude regime to the aerosol optical depth is estimated. 5-year ( ) mean AOD and km integrated extinction are indicated by solid lines in Figure 8. Five-year mean 550-nm AOD is found to be 0.32 while the 5-year 10- to 30-km integrated extinction at 532 nm from lidar is estimated to be This value indicates that on an average about 12% of AOD is contributed from 10 to 30 km while the rest comes from the lower altitudes of the troposphere. AODs derived from MODIS are found to be higher than 0.4 during March November 2004 and 2005 respectively (Figure 8a). Both the columnar AOD and the 10- to 30-km integrated extinction are found to exhibit seasonal variations suggesting that the contribution from 10 to 30 km to the columnar AOD can vary. For example, the 10- to 30-km integrated extinction is about 0.04 or higher during 2001, while MODIS AODs are in the range; this corresponds to a contribution of 10 20% from 10 to 30 km to the columnar AOD. It should also be noted that AODs over Gadanki on an average are higher during 2004 and 2005 when compared to the previous years. A similar trend of higher AODs during in 2004 and 2005 when compared to has been found in other locations in south India [Ramachandran and Cherian, 2008]. 9of11

10 [33] The relation between the columnar AODs and the integrated extinction in the 10- to 30-km region is further explored. AODs over Gadanki shows a summer high and winter low; winter (DJF) mean AOD is 0.27 ± 0.03 and summer (JJA) mean AOD over Gadanki during is 0.37 ± The magnitude of summer increase in AOD with respect to winter is about 1.4. This lower summer to winter AOD ratio over Gadanki is quite consistent with the value obtained over nearby Chennai (13.0 N, 80.2 E) [Ramachandran, 2007]. The monthly mean AODs and integrated extinction from 10 to 30 km during yielded a positive correlation coefficient of 0.62, suggesting that aerosols in the lower troposphere could contribute to UT and LS aerosols over Gadanki. The increasing trends in UT and LS aerosols, the yearly increase in AOD during and the lower-upper troposphere coupling over Gadanki are relevant in terms of climate change over the Indian subcontinent, as the increasing aerosol abundance over India may be impacting the monsoon [Ramanathan et al., 2005]. 5. Summary and Conclusions [34] Seasonal and altitudinal variations of aerosol characteristics measured over Gadanki (13.5 N, 79.2 E), a tropical station in south India, using a 532-nm Nd:YAG lidar during are studied. The lidar was operated only during clear sky conditions. The daily average aerosol profiles are constructed from about 5 hours of observation each night. The profiles obtained have a range resolution of 300 m and a time resolution of 250 s. Results obtained from about 400 days of observations from January 2001 to December 2005 are analyzed and reported. [35] Aerosol scattering ratio and extinction coefficient in the 10- to 30-km altitude region are derived using Klett top to bottom integration algorithm. The occurrence of cirrus clouds over Gadanki is seen to hinder aerosol profile measurements. The frequency of occurrence of cirrus clouds in the 9- to 19-km altitude region over Gadanki during peaks at 22% around the tropopause region of km. Scattering ratios (SR) did not exceed 1.2 on non-cirrus days in the upper troposphere lower stratosphere region over Gadanki. On the basis of this a threshold value of 1.25 for SR was chosen for analyzing the lidar data over Gadanki to separate aerosols and cirrus clouds. As there has been no major volcanic eruption since June 1991 (Mt. Pinatubo) and as the residence time of volcanic aerosols is about three years, the aerosol characteristics discussed in this study pertain to volcanically quiescent conditions. [36] SR and aerosol extinction are found to exhibit large seasonal and interannual variations in the upper troposphere (UT, km) and lower stratosphere (LS, km). SR was found to be about throughout the year in 2001 except in October November. In 2004 SR was low in the 10- to 15-km region with values of about 1.1 during January March. Aerosol extinctions are found to show similar variability; extinctions in the 10- to 13-km altitude range are about km 1. Higher SR values are seen in the 18- to 20-km region during the year, with incursions of higher SR up to about 25 km during February and December In 2004 an increase in SR values is seen during March April May and October November up to the altitude region of 25 km. Extinction is found to be about km 1 in the 18- to 20-km region and decreases sharply there after. [37] SR in UT in 2001 and 2004 during winter (DJF) is lower than that of summer (JJA), while in LS winter profiles are found to have higher SR values. In 2001 and 2004 the mean SR values are higher during winter than in summer in LS. In 2005 no significant winter summer differences in SR are seen in UT and LS which could be due to minor volcanic eruption that took place in Over Gadanki the tropopause is colder in winter as compared to summer. A general decrease in SR is seen during both winter and summer above tropopause. SR values are found to apparently experience a shift in altitude corresponding to the seasonal change in cold point tropopause indicating a relation between SR and tropopause altitude. [38] Time evolution of aerosol extinctions integrated in UT (10 15 km) and LS (18 30 km) region over Gadanki during showed increasing trends of 5 and 17% respectively, while long-term global aerosol data showed no significant change. The increasing trend in UT extinction from lidar is consistent with the earlier results obtained over India using SAGE II data [Ramachandran and Jayaraman, 2003]. The increasing trends in UT and LS aerosols over Gadanki seem to corroborate the finding that emissions from subtropical and tropical Asia may have already started to influence the amount of sulfur containing gases reaching the stratosphere. It has been noted that OCS in the upper tropical troposphere may also have increased due to increases in biomass burning. The increase in UT aerosols is important as the LS aerosols depend on inputs from the upper troposphere. [39] A scatter of monthly mean UT and LS aerosol extinctions over Gadanki resulted in a negative correlation coefficient of 0.4. The study period was marked by minor volcanic eruptions of Reventador (0.077 S, W, 3 November 2002) and Manam (4.10 S, E, January 2005) which could have resulted in increased SR values during November 2002 January 2003 and February A strong coupling between UT and LS aerosols was observed as LS aerosols were found to be the predominating source over Pasadena. The UT and LS integrated extinctions were comparable resulting in a better correlation over Pasadena, while in Gadanki UT aerosol extinctions are higher suggesting that UT aerosols perhaps could be a source of LS aerosols. [40] The lidar integrated aerosol extinction in the 10- to 30-km altitude region was found to contribute about 12% to the aerosol optical depth (AOD) derived using MODIS. The relation between monthly mean AOD and the integrated extinction obtained from the 5-year data yielded a correlation coefficient of 0.62 suggesting that aerosols in the lower troposphere could contribute to UT aerosols over Gadanki. The increases in UT and LS aerosols during and the correlation between UT and LS aerosols over Gadanki are important in terms of climate change over India as it has been shown that increasing aerosol concentrations may impact the monsoon. 10 of 11

11 References Barnes, J. E., and D. J. Hofmann (2001), Variability in the stratospheric background aerosol over Mauna Loa Observatory, Geophys. Res. Lett., 28, Bekki, S., and J. A. Pyle (1994), Two-dimensional modeling study of the volcanic eruption of Mount Pinatubo, J. Geophys. Res., 99, 18,861 18,869. Brock, C. A., P. Hamill, J. C. Wilson, H. H. Jonsson, and K. R. Chan (1995), Particle formation in the upper troposphere: A source of nuclei for the stratospheric aerosol, Science, 270, Fernald, F. G. (1984), Analysis of atmospheric lidar observations: Some comments, Appl. Opt., 23, Hamill, P., E. J. Jensen, P. B. Russell, and J. J. Bauman (1997), The life cycle of stratospheric aerosol particles, Bull. Am. Meteorol. Soc., 78, Hervig, M., and T. Deshler (2002), Evaluation of aerosol measurements from SAGE II, HALOE, and balloonborne optical particle counters, J. Geophys. Res., 107(D3), 4031, doi: /2001jd Hitchman, H., M. McKay, and C. R. Trepte (1994), A climatology of stratospheric aerosol, J. Geophys. Res., 99, 20,689 20,700. Hofmann, D. J. (1993), Twenty years of balloon-borne tropospheric aerosol measurements at Laramie, Wyoming, J. Geophys. Res., 98, 12,753 12,766. Intergovernmental Panel on Climate Change IPCC (2001), Climate Change 2001: The Scientific Basis, edited by J. T. Houghton et al., 881 pp., Cambridge Univ. Press, Cambridge, U.K. Intergovernmental Panel on Climate Change IPCC (1999), Aviation and the Global Atmosphere, edited by J. E. Penner et al., 373 pp., Cambridge Univ. Press, Cambridge, U.K. Jäger, H., and D. J. Hofmann (1991), Midlatitude lidar backscatter to mass, area, and extinction conversion model based on in situ aerosol measurements from 1980 to 1987, Appl. Opt., 30, Jäger, H., O. Uchino, T. Nagai, T. Fujimoto, V. Freudenthaler, and F. Homburg (1995), Ground-based remote sensing of the decay of the Pinatubo eruption cloud at three northern hemisphere sites, Geophys. Res. Lett., 22, Jayaraman, A., S. Ramachandran, Y. B. Acharya, and B. H. Subbaraya (1995), Pinatubo volcanic aerosol layer decay observed at Ahmedabad (23 N), India, using neodymium:yttrium/aluminium/garnet backscatter lidar, J. Geophys. Res., 100, 23,209 23,214. Kamei, A., N. Sugimoto, I. Matsui, A. Shimizu, and T. Shibata (2006), Volcanic aerosol layer observed by shipboard lidar over the tropical western pacific, SOLA, 2, 1 4, doi: /sola Kent, G. S., C. R. Trepte, and P. L. Lucker (1998), Long-term stratospheric aerosol and gas experiment I and II measurements of upper tropospheric aerosol characteristics, J. Geophys. Res., 103, 28,863 28,874. Klett, J. D. (1985), Lidar inversion with variable backscatter/extinction ratios, Appl. Opt., 24, Menzies, R. T., and D. M. Tratt (1995), Evidence of seasonally dependent stratosphere troposphere exchange and purging of lower stratospheric aerosol from a multiyear lidar data set, J. Geophys. Res., 100, Notholt, J., et al. (2003), Enhanced upper tropical tropospheric COS: Impact on the stratospheric aerosol layer, Science, 300, Notholt, J., et al. (2005), Influence of tropospheric SO 2 emissions on particle formation and the stratospheric humidity, Geophys. Res. Lett., 32, L07810, doi: /2004gl Osborn, M. T., R. J. DeCoursey, C. R. Trepte, D. M. Winker, and D. C. Woods (1995), Evolution of the Pinatubo volcanic cloud over Hampton, Virginia, Geophys. Res. Lett., 22, Parameswaran, K., S. V. Sunilkumar, B. V. Krishna Murthy, and K. Satheesan (2004), Lidar observations of high altitude cirrus clouds near the tropical tropopause, Adv. Space Res., doi: /j.asr Ramachandran, S. (2007), Aerosol optical depth and fine mode fraction variations deduced from Moderate Resolution Imaging Spectroradiometer (MODIS) over four urban areas in India, J. Geophys. Res., 112, D16207, doi: /2007jd Ramachandran, S., and R. Cherian (2008), Regional and seasonal variations in aerosol optical characteristics and their frequency distributions over India during , J. Geophys. Res., 113, D08207, doi: / 2007JD Ramachandran, S., and A. Jayaraman (2003), Balloon-borne study of the upper tropospheric and stratospheric aerosols over a tropical station in India, Tellus, 55B, Ramanathan, V., et al. (2005), Atmospheric brown clouds: Impacts on south Asian climate and hydrological cycle, Proc. Natl. Acad. Sci., 102, Reischmann, T., T. Toulkeridis, and E. Aguilera (2003), Re-Awakening of a volcano: The 3, November 2002 Eruption of El Reventador, Ne Ecuador, Geophys. Res. Abstr., 5, Remer, L. A., et al. (2005), The MODIS algorithm, products and validation, J. Atmos. Sci., 62, Revathy, K., S. R. P. Nayar, and B. V. Krishna Murthy (1996), Deduction of temperature profile from MST radar observations of vertical wind, Geophys. Res. Lett., 23, Rosen, J. M., and D. J. Hofmann (1975), Stratospheric aerosol measurements: II. The world wide distribution, J. Atmos. Sci., 32, Rosenlof, K. H. (1995), Seasonal cycle of the residual mean meridional circulation in the stratosphere, J. Geophys. Res., 100, Satheesan, K., and B. V. Krishna Murthy (2005), A study of tropical tropopause using MST radar, Ann. Geophys., 23, Seinfeld, J. H., and S. N. Pandis (1998), Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, 1326 pp., John Wiley Sivakumar, V., Y. Bhavanikumar, P. B. Rao, K. Mizutani, T. Aoki, M. Yasui, and T. Itabe (2003), Lidar observed characteristics of the tropical cirrus clouds, Radio Sci., 38(6), 1094, doi: /2002rs SPARC (2006), Assessment of Stratospheric Aerosol Properties, WCRP- 124, WMO/TD - No. 1295, SPARC Rep. No. 4, edited by L. W. Thomason and Th. Peter, 348 pp. Sunilkumar, S. V., and K. Parameswaran (2005), Temperature Dependence of Tropical Cirrus Properties and Radiative effects, J. Geophys. Res., 110, D13205, doi: /2004jd Thomason, L. W., and G. Taha (2003), SAGE III aerosol extinction measurements: Initial results, Geophys. Res. Lett., 30(12), 1631, doi: / 2003GL Trepte, C. R., L. W. Thomason, and G. S. Kent (1994), Banded structures in stratospheric aerosols distributions, Geophys. Res. Lett., 21, Uchino, O., T. Nagai, T. Fujjimoto, W. A. Matthews, and J. Orange (1995), Extensive Lidar observations of the Pinatubo aerosol layers at Tsukuba (36.1 N), Japan and Lauder (45.0 S), New Zealand, Geophys. Res. Lett., 22, U.S. Standard Atmosphere Supplements (1966), NASA, 289 pp., U.S. Government Printing Office, Washington D.C. Y. Bhavani Kumar, P. Kulkarni, and D. Narayana Rao, Department of Space, National Atmospheric Research Laboratory, Gadanki , India. (ypbk@narl.gov.in; padma.narl@gmail.com; profdnrao2001@yahoo.com) M. Krishnaiah, Department of Physics, S.V. University, Tirupati , India. (profkrishnaiah.m@gmail.com) S. Ramachandran, Space and Atmospheric Sciences Division, Physical Research Laboratory, Ahmedabad , India. (ram@prl.res.in) 11 of 11