**4. Altitude structure of backscatter/extinction, scattering ratio, and depolarization from lidar and comparison with SAGE measurements**

Figure 1 shows the altitude profiles of mean particulate backscatter coefficient (βp), effective backscatter ratio (R) and volume depolarization ratio (δ) for a few nights at Gadanki obtained from Lidar data during the year 1999 as typical samples. In general, βp and R show a general decrease with increase in altitude (eg. 05 and 12 April 1999) in the troposphere and stratosphere. But on a few nights a significant enhancement is observable over a small region between 9 and 17 km. This sharp increase in βp is due to strong scattering from ice particles of thin STC layers, formed at these altitudes either by *in situ* condensation of water vapour or originated from the outflow of convective anvils [43-45]. On a few occasions this layer of enhanced βp extends down, up to 5–6 km, depicting typical case of dense cirrus (9 June 1999), occurring predominantly during the monsoon period. Note that, these STCs are so transparent that the lidar beam could penetrate the cloud and provide measurable signal even from higher altitudes. The opacity of STCs are quantified using the cloud optical depth (τc) which is the height integrated particulate extinction coefficient (αp), obtained by multiplying βp with the Sp, from the cloud base (hcb) up to the cloud top (hct). In case if the cirrus is too dense (with τc exceeding 1.5) the lidar beam will not be able to penetrate the cirrus layer impeding useful lidar observations. In association with the enhancement in β<sup>p</sup> and R, a significant increase in δ also can be observed at these altitudes. This suggests that the scattering particles within the STCs are relatively large and significantly non-spherical in nature. Depending on τc, STCs are further classified [46] in three classes viz., sub-visual cirrus (SVC) with τc<0.03, thin cirrus (TC) with 0.03 < τc <0.3 and dense cirrus (DC) with τ<sup>c</sup> >0.3. General features of STCs from this tropical station [37] showed that while the occurrence of SVC is larger during winter, TC and DC occur more frequently during the monsoon period. The upper and lower boundaries of STCs are identified from altitude profiles of R and δ using a threshold condition [37,42] for R to exceed 2 in either of the two lidar channels (⊥ or ‖ channels) along with the value of δ exceeding 0.04 in the altitude region where the STCs are usually observed ( 8 to 20 km).

116 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

altitude profile of βp with the corresponding profile of SP.

**4. Altitude structure of backscatter/extinction, scattering ratio, and depolarization from lidar and comparison with SAGE measurements** 

Figure 1 shows the altitude profiles of mean particulate backscatter coefficient (βp), effective backscatter ratio (R) and volume depolarization ratio (δ) for a few nights at Gadanki obtained from Lidar data during the year 1999 as typical samples. In general, βp and R show a general decrease with increase in altitude (eg. 05 and 12 April 1999) in the troposphere and stratosphere. But on a few nights a significant enhancement is observable over a small region between 9 and 17 km. This sharp increase in βp is due to strong scattering from ice particles of thin STC layers, formed at these altitudes either by *in situ* condensation of water vapour or originated from the outflow of convective anvils [43-45]. On a few occasions this layer of enhanced βp extends down, up to 5–6 km, depicting typical case of dense cirrus (9 June 1999), occurring predominantly during the monsoon period. Note that, these STCs are

backscattering coefficients (β⊥ and β║, respectively) taking 30 km as the reference altitude where the aerosol contribution is assumed to be negligible. For this inversion a value of 40 sr-1 is assigned for the lidar ratio (SP) and its variation with altitude depending on δ is also accounted appropriately. With this correction the value of SP reduces to ∼26 sr-1 within the STC [39] in the upper troposphere. Further incorporating the correction for multiple scattering the value of SP within the STC reduces to 20 sr-1 (which is used to study the properties of STCs). The molecular backscatter coefficients for the two polarized components are estimated from the mean molecular number density profile taking a molecular depolarization factor (δm) of 0.028 [41]. The molecular backscatter coefficient of the co-polarized component (β<sup>m</sup>⊥) is related to that of the cross-polarized component (β<sup>m</sup>║) as β<sup>m</sup>⊥ =δmβ<sup>m</sup>║. Subtracting β<sup>m</sup>⊥ and β<sup>m</sup>║ from the altitude profiles of β⊥ and β║, respectively obtained from lidar data employing the Fernald's algorithm, the altitude profiles of particulate backscatter coefficient, β<sup>p</sup>⊥ and β<sup>p</sup>║, are estimated. The respective backscatter ratios for the co-polarized (R⊥) and cross-polarized (R║) components are estimated [37] as R⊥=β⊥/β<sup>m</sup>⊥ and R║=β║/β<sup>m</sup>║. As far the net atmospheric backscattering is concerned, the ''unbiased'' or effective backscatter ratio (R) is to be defined , to quantify the gross property of the medium, which on mathematical simplification can be written as R(h)=[R⊥(h)+δmR║(h)]/(1+δm). The volume depolarization ratio is obtained from the ratios of R⊥ and R║ as δ(h)= [δm R║(h)]/ R⊥(h). This ratio is a good indicator for distinguishing the cirrus based on 'particle habit'. While for small spherical particles, the values of δ will be relatively small, its value increases significantly as they become large and non- spherical. Using this property of cloud particles, structure and altitude extent of cirrus can be estimated from each lidar profile, which will be used to study the temporal variation of cirrus properties during the entire period of lidar observation. Based on a detailed scrutiny of a number of profiles at different cloud conditions a threshold value of δ ≥0.04 is assigned for discriminating the STC [42]. If the value of δ exceeds this threshold value it is classified as STC. The effective particulate backscatter coefficient, βP, is the sum of β<sup>p</sup>⊥ and β<sup>p</sup>║. The altitude profile of particulate extinction coefficient, αP, is estimated by multiplying the

**Figure 1.** Altitude profiles of mean particulate backscatter coefficient (βp), effective backscatter ratio (R) and volume depolarization ratio (δ) for few nights during the year 1999

A detailed error analysis [42] showed that the estimated values of βp is less sensitive to the variability in SP. For a given uncertainty of 25% in Sp, the maximum uncertainty in βp is 10% in the absence of clouds, ~15% for thin cirrus and ~30% for thick cirrus. For the same uncertainty in SP, the maximum uncertainty in the retrieved backscatter coefficient and effective backscatter ratio are around 0.6%, 2% and 10%, respectively, for clear atmosphere, atmosphere with thin cirrus and atmosphere with thick cirrus. Including the possible errors in the lidar signal inversion associated with the uncertainty in the molecular backscatter coefficient, the resultant error in the derived optical depth would be ~20%. As the signal-tonoise ratio is >2 up to ∼45 km, for altitudes <30 km the system induced errors will be significantly small (<1%) compared to that from other sources. The error due to the influence of background and system noise is negligible (< 0.001%) compared to that due to the uncertainty in Sp.

Distribution of Particulates in the Tropical UTLS over

the Asian Summer Monsoon Region and Its Association with Atmospheric Dynamics 119

extinction measured by the two instruments in the altitude range 18–25 km is <40%, which is comparable in magnitude with those obtained in other similar inter-comparisons [48-50]. For Altitudes <17 and >25 km, both mean differences and standard deviations are relatively large (< 40%). This could partly be due to the temporal variations during the course of the two measurements as well as the spatial heterogeneity (between the locations of the two measurements). The observed increase in deviation below 17 km is partly due to the

Space-borne lidars, *in situ* measurements and ground-based experiments indicates frequent manifestation of cirrus clouds in the upper troposphere [36,51-53]. The frequency of occurrence of STC (FSTC) over a wide geographical region can be derived from VHRR data from remote sensing satellites. The geostationary meteorological satellite, KALPANA-1, positioned at 74°E over the equator for continuous measurements of clouds and convective systems, provides the required information over the Indian region. This satellite observes the earth in three wavelength bands: Visible (0.55–0.75 μm), Water vapor band (WV: 5.7–7.1 μm), and the atmospheric window of thermal infrared (TIR: 10.5–12.5 μm).. In TIR and WV bands, the data is recorded at a pixel resolution of 8 km (nadir) with a digital resolution of 10 bits. Unless the cloud is optically thick, the radiance observed in TIR band does not correspond only to the cloud top, but is also weighted by the radiation emitted from the altitudes below. The brightness temperatures measured in these two channels are used to detect STC following the bi-spectral approach of Roca et al. [54]. In this method all the cloudy pixels having WV brightness temperature < 246 K and TIR brightness temperature > 270 K are treated as STC. In addition to the above, those cloudy pixels having a brightness temperature difference > 20 K between the two channels and having WV brightness temperature < 246 K, a condition imposed mainly to detect STC above other low level clouds, also are treated as STC [55]. Figure 3 shows the frequency of occurrence of STC (FSTC) over the Indian region in different months for the year 2005 derived from KALPANA-1 data. Any thin cirrus cloud above a deep convective cloud cannot be detected by the present bi-spectral algorithm, unless the difference in brightness temperature in the two channels exceeds 20 K (indicating significant altitudinal separation between the top of the STC and the optically dense high-altitude cloud), which may not be the case if STC forms just above the high altitude cloud. Because of this inherent limitation, the estimated FSTC will always be an underestimate over the region where high-altitude clouds are present. Hence the regions where the monthly mean frequency of occurrence of high-altitude clouds larger than 20%

For studying the role of deep convection on the genesis of STC, the monthly mean spatial distribution of the frequency of occurrence of deep clouds (FD) with TIR brightness temperature < 235 K derived from KALPANA-1 during January-December 2005 are presented in Figure 4. A comparison of Figures 3 and 4 suggests that that the longitudinally extended band of high FSTC (~40-60%) in the region 10°S-20°S over the western tropical Indian Ocean and around the equator over the eastern Indian Ocean in January (Figure 3) is

influence of STCs in the UT region as well as their spatial heterogeneity.

**5. Semitransparent cirrus in the upper troposphere** 

are masked (dark brown) in Figure 3.

Altitude profiles of αp at 525 and 1020 nm from Global SAGE-II aerosol data archive (version 6.2) during the period 1998-2005 are obtained through the NASA website *http://wwwsage2.larc.nasa.gov/data/v6\_data*. Typical estimated error in SAGE-II measured αp at 525 and 1020 nm are in the range 10 to 15% [34]. Figure 2 show a comparison of αp obtained from the SAGE-II at 525 and 1020 nm along with that of the lidar at 532 nm in the altitude region 10–30 km for a few sunset occultation events during the period 1998–2003. These comparisons are made when SAGE-II had an occultation pass within a grid size of ± 5° in latitude and ±10° in longitude centered at Gadanki within a time-duration of 1 day with respect to the lidar observation. As the difference between the SAGE-II wavelength of 525 nm and lidar wavelength of 532 nm is less than 1.5%, the expected absolute differences in α<sup>P</sup> for these two wavelengths would be almost insignificant [47]. The latitude and longitude of line-of-sight tangent point of SAGE-II for each occultation event are shown in the respective frames of Figure 2. The radial distance, d, between the tangent point and the lidar location estimated from the latitudinal and longitudinal differences between the two is also marked in this figure. In general, the shape of the SAGE-II and lidar-derived extinction profiles show a good agreement especially in the stratosphere. The mean percentage difference for

**Figure 2.** A comparison of the altitude profiles of αp on a few days derived from lidar data (532nm) at Gadanki, along with that of SAGE-II (525 and 1020 nm) sunset occultation events near this region.

extinction measured by the two instruments in the altitude range 18–25 km is <40%, which is comparable in magnitude with those obtained in other similar inter-comparisons [48-50]. For Altitudes <17 and >25 km, both mean differences and standard deviations are relatively large (< 40%). This could partly be due to the temporal variations during the course of the two measurements as well as the spatial heterogeneity (between the locations of the two measurements). The observed increase in deviation below 17 km is partly due to the influence of STCs in the UT region as well as their spatial heterogeneity.

## **5. Semitransparent cirrus in the upper troposphere**

118 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

uncertainty in Sp.

of background and system noise is negligible (< 0.001%) compared to that due to the

Altitude profiles of αp at 525 and 1020 nm from Global SAGE-II aerosol data archive (version 6.2) during the period 1998-2005 are obtained through the NASA website *http://wwwsage2.larc.nasa.gov/data/v6\_data*. Typical estimated error in SAGE-II measured αp at 525 and 1020 nm are in the range 10 to 15% [34]. Figure 2 show a comparison of αp obtained from the SAGE-II at 525 and 1020 nm along with that of the lidar at 532 nm in the altitude region 10–30 km for a few sunset occultation events during the period 1998–2003. These comparisons are made when SAGE-II had an occultation pass within a grid size of ± 5° in latitude and ±10° in longitude centered at Gadanki within a time-duration of 1 day with respect to the lidar observation. As the difference between the SAGE-II wavelength of 525 nm and lidar wavelength of 532 nm is less than 1.5%, the expected absolute differences in α<sup>P</sup> for these two wavelengths would be almost insignificant [47]. The latitude and longitude of line-of-sight tangent point of SAGE-II for each occultation event are shown in the respective frames of Figure 2. The radial distance, d, between the tangent point and the lidar location estimated from the latitudinal and longitudinal differences between the two is also marked in this figure. In general, the shape of the SAGE-II and lidar-derived extinction profiles show a good agreement especially in the stratosphere. The mean percentage difference for

**Figure 2.** A comparison of the altitude profiles of αp on a few days derived from lidar data (532nm) at Gadanki, along with that of SAGE-II (525 and 1020 nm) sunset occultation events near this region.

Space-borne lidars, *in situ* measurements and ground-based experiments indicates frequent manifestation of cirrus clouds in the upper troposphere [36,51-53]. The frequency of occurrence of STC (FSTC) over a wide geographical region can be derived from VHRR data from remote sensing satellites. The geostationary meteorological satellite, KALPANA-1, positioned at 74°E over the equator for continuous measurements of clouds and convective systems, provides the required information over the Indian region. This satellite observes the earth in three wavelength bands: Visible (0.55–0.75 μm), Water vapor band (WV: 5.7–7.1 μm), and the atmospheric window of thermal infrared (TIR: 10.5–12.5 μm).. In TIR and WV bands, the data is recorded at a pixel resolution of 8 km (nadir) with a digital resolution of 10 bits. Unless the cloud is optically thick, the radiance observed in TIR band does not correspond only to the cloud top, but is also weighted by the radiation emitted from the altitudes below. The brightness temperatures measured in these two channels are used to detect STC following the bi-spectral approach of Roca et al. [54]. In this method all the cloudy pixels having WV brightness temperature < 246 K and TIR brightness temperature > 270 K are treated as STC. In addition to the above, those cloudy pixels having a brightness temperature difference > 20 K between the two channels and having WV brightness temperature < 246 K, a condition imposed mainly to detect STC above other low level clouds, also are treated as STC [55]. Figure 3 shows the frequency of occurrence of STC (FSTC) over the Indian region in different months for the year 2005 derived from KALPANA-1 data. Any thin cirrus cloud above a deep convective cloud cannot be detected by the present bi-spectral algorithm, unless the difference in brightness temperature in the two channels exceeds 20 K (indicating significant altitudinal separation between the top of the STC and the optically dense high-altitude cloud), which may not be the case if STC forms just above the high altitude cloud. Because of this inherent limitation, the estimated FSTC will always be an underestimate over the region where high-altitude clouds are present. Hence the regions where the monthly mean frequency of occurrence of high-altitude clouds larger than 20% are masked (dark brown) in Figure 3.

For studying the role of deep convection on the genesis of STC, the monthly mean spatial distribution of the frequency of occurrence of deep clouds (FD) with TIR brightness temperature < 235 K derived from KALPANA-1 during January-December 2005 are presented in Figure 4. A comparison of Figures 3 and 4 suggests that that the longitudinally extended band of high FSTC (~40-60%) in the region 10°S-20°S over the western tropical Indian Ocean and around the equator over the eastern Indian Ocean in January (Figure 3) is

closely associated with the deep convection linked to the Inter tropical convergence Zone. However, the values of FSTC far exceed FD over all these regions. During this month, the highest values of FSTC (~60%) are observed at the north of Madagascar in the western Indian Ocean and over Sumatra/Indonesia in the eastern Indian Ocean, where the frequency of occurrence of very large deep convection is quite large. A region of less cloudiness (with FSTC<15%) is observed in the central Arabian Sea and Indian Peninsula centered around 5°N-15°N, which runs parallel to the equatorial band of high STC occurrence and is well separated from the deep convective regions. A similar STC-free zone is also observed in the southeast Indian Ocean centered on 20°S. Clearly, these regions with low occurrence of STC are caused by the large subsidence in the upper troposphere associated with the descending limb of the Hadley circulation cell.

Distribution of Particulates in the Tropical UTLS over

the Asian Summer Monsoon Region and Its Association with Atmospheric Dynamics 121

**Figure 4.** Spatial variation of deep convective clouds in different months during the year 2005 derived

**Figure 5.** Month-to-month variation of the normalized frequency of encountering STC at different attitudes in the upper troposphere (a) along with the month-to-month variation of the mean top, base and optic center of STC with vertical bars indicating the standard error (b) from lidar data at Gadanki

from the KALPANA-1VHRR data

for the period 1998–2003.

**Figure 3.** Spatial variation of mean STC occurrence in different months during the year 2005 (brown color shows the region where the STC retrieval was not possible due to the presence of high clouds) derived from the KALPANA-1 VHRR data

## **6. General features of STC over the Indian region**

Lidar studies from Gadanki indicate that the occurrence of STC in this geographical region is the largest from May-October, associated with the formation of intense convection and the subsequent onset of Asian summer monsoon (ASM). Figure 5 shows the mean feature of STC occurrence and its altitude extend averaged for the period 1998-2003. While the STCs

limb of the Hadley circulation cell.

derived from the KALPANA-1 VHRR data

**6. General features of STC over the Indian region** 

closely associated with the deep convection linked to the Inter tropical convergence Zone. However, the values of FSTC far exceed FD over all these regions. During this month, the highest values of FSTC (~60%) are observed at the north of Madagascar in the western Indian Ocean and over Sumatra/Indonesia in the eastern Indian Ocean, where the frequency of occurrence of very large deep convection is quite large. A region of less cloudiness (with FSTC<15%) is observed in the central Arabian Sea and Indian Peninsula centered around 5°N-15°N, which runs parallel to the equatorial band of high STC occurrence and is well separated from the deep convective regions. A similar STC-free zone is also observed in the southeast Indian Ocean centered on 20°S. Clearly, these regions with low occurrence of STC are caused by the large subsidence in the upper troposphere associated with the descending

**Figure 3.** Spatial variation of mean STC occurrence in different months during the year 2005 (brown color shows the region where the STC retrieval was not possible due to the presence of high clouds)

Lidar studies from Gadanki indicate that the occurrence of STC in this geographical region is the largest from May-October, associated with the formation of intense convection and the subsequent onset of Asian summer monsoon (ASM). Figure 5 shows the mean feature of STC occurrence and its altitude extend averaged for the period 1998-2003. While the STCs

**Figure 4.** Spatial variation of deep convective clouds in different months during the year 2005 derived from the KALPANA-1VHRR data

**Figure 5.** Month-to-month variation of the normalized frequency of encountering STC at different attitudes in the upper troposphere (a) along with the month-to-month variation of the mean top, base and optic center of STC with vertical bars indicating the standard error (b) from lidar data at Gadanki for the period 1998–2003.

occurring during the period May- October are generally thick and optically denser [37], those occurring during the rest of the year (dry months) are relatively thin (both geometrically and optically). As the Gadanki region is almost free from deep convection (Figure 4) during winter, the STCs forming during this period could be of *in situ* origin. While the dense STCs observed during the May-October period are associated with deep convection over the Indian land mass and Bay of Bengal.

Distribution of Particulates in the Tropical UTLS over

the Asian Summer Monsoon Region and Its Association with Atmospheric Dynamics 123

**Figure 7.** Mean variation of Cloud thickness, Volume Depolarization Ratio and cloud optical depth

**7. Mean annual variation of particulate scattering in the UTLS region** 

Figure 8a shows a contour plot of the logarithm of mean βp as a function of month and altitude for the period 1998–2003. The values of βp are relatively large during the May to September period and small during the winter months. This is due to the influence of STC. Relatively high values of βp in the UT region during the monsoon period are due to the presence of relatively dense cirrus and low values during winter are due to the presence of SVC. In contrast to the UT region, βp in the LS is generally large (as high as 6 × 10-8 m-1 sr-1) during the winter (November to January) and pre-monsoon (April–May) months and low (as low as 10-9 m-1 sr-1) during the summer (July and August) months. Prominent peaks are observed during May–June and November–January periods with low values in July–August and February. The mean annual pattern of αp at different altitudes from the lidar derived β<sup>p</sup> (Figure8a) is presented in Figure 8b. Similar grading scheme is used in both these plots to make an easy direct visual comparison of the pattern. The major features of the annual variation of βp and αp in different altitudes are very similar in these two plots, even though the lidar ratio is assumed to be variable with altitude depending on aerosol properties.

Microphysical properties like the size and shape of particles in the UTLS region can be delineated from the depolarization of backscattered radiation [57,58]. Figure 8c shows a contour plot of monthly mean δ in the altitude region 8–28 km. The value of δ varies in the range 0.03 to 0.6 in UT region and from 0.03– 0.04 in the LS region. High values of δ are

temperature. The value of δ also shows a sharp decrease when the cloud temperature decreases below −75°C. The decrease in δ with increase in temperature can be attributed to the melting and evaporation of ice crystals and subsequent blunting of their edges [56]. Decrease in temperature leads formation of particles with sharp edges. But when the temperature decreases below a threshold value, the particle size [56] becomes small (needle type). This can lead to a decrease in δ. Figure 7c shows that on an average τc increases with increase in temperature. The cloud becomes more opaque at higher temperatures. This has important implication in the radiative effects of STCs [38] in the context of global change.

with mid-cloud temperature for STCs observed at Gadanki during the period 1998-2003.

On examining the frequency of occurrence of STC during the period 1998-2003 (Figure 6a) it can be seen that the frequency of occurrence of SVC is much larger than TC and DC. In general, STCs are observed in the altitude region 10 to 18 km with a preferred altitude (frequent occurrence) region between 14 and 16 km (Figure 6b) Thin clouds occur more frequently than thick clouds (Figure 6c). Though the vertical extent of STC generally vary from 0.4 to ~ 4.0 km in majority of the cases it is less than 1.7 km. Though the volume depolarization,δ, in these clouds varies in the range 0.03 to 0.6 its distribution peaks (Figure 6d) in the lowest value. The value of δ for SVC and TC are generally very small compared to DC. The particulate depolarization (δp) of STC (Figure 6e) generally varies from zero to unity. The distribution of δp peaks around a value of 0.15. The properties of STC vary significantly with cloud temperature (or altitude). Figure 7 shows the variation of thickness, depolarization and optical depth of STC with cloud temperature. As can be seen from Figure 7a, the thickness of STC is a maximum for temperatures in the range −55° to −75°C. Above and below this temperature range the cloud thickness decreases. Similarly the depolarization is maximum around −75°C and decreases steadily with increase in

**Figure 6.** Frequency distribution of (a) cloud optical depth, (b) mid-cloud altitude (optic center), (c) Geometrical thickness, (e) Volume Depolarization Ratio, (f) Particulate depolarization, and (g) Midcloud temperature of STCs observed at Gadanki for the period 1998-2003.

convection over the Indian land mass and Bay of Bengal.

occurring during the period May- October are generally thick and optically denser [37], those occurring during the rest of the year (dry months) are relatively thin (both geometrically and optically). As the Gadanki region is almost free from deep convection (Figure 4) during winter, the STCs forming during this period could be of *in situ* origin. While the dense STCs observed during the May-October period are associated with deep

On examining the frequency of occurrence of STC during the period 1998-2003 (Figure 6a) it can be seen that the frequency of occurrence of SVC is much larger than TC and DC. In general, STCs are observed in the altitude region 10 to 18 km with a preferred altitude (frequent occurrence) region between 14 and 16 km (Figure 6b) Thin clouds occur more frequently than thick clouds (Figure 6c). Though the vertical extent of STC generally vary from 0.4 to ~ 4.0 km in majority of the cases it is less than 1.7 km. Though the volume depolarization,δ, in these clouds varies in the range 0.03 to 0.6 its distribution peaks (Figure 6d) in the lowest value. The value of δ for SVC and TC are generally very small compared to DC. The particulate depolarization (δp) of STC (Figure 6e) generally varies from zero to unity. The distribution of δp peaks around a value of 0.15. The properties of STC vary significantly with cloud temperature (or altitude). Figure 7 shows the variation of thickness, depolarization and optical depth of STC with cloud temperature. As can be seen from Figure 7a, the thickness of STC is a maximum for temperatures in the range −55° to −75°C. Above and below this temperature range the cloud thickness decreases. Similarly the depolarization is maximum around −75°C and decreases steadily with increase in

**Figure 6.** Frequency distribution of (a) cloud optical depth, (b) mid-cloud altitude (optic center), (c) Geometrical thickness, (e) Volume Depolarization Ratio, (f) Particulate depolarization, and (g) Mid-

cloud temperature of STCs observed at Gadanki for the period 1998-2003.

**Figure 7.** Mean variation of Cloud thickness, Volume Depolarization Ratio and cloud optical depth with mid-cloud temperature for STCs observed at Gadanki during the period 1998-2003.

temperature. The value of δ also shows a sharp decrease when the cloud temperature decreases below −75°C. The decrease in δ with increase in temperature can be attributed to the melting and evaporation of ice crystals and subsequent blunting of their edges [56]. Decrease in temperature leads formation of particles with sharp edges. But when the temperature decreases below a threshold value, the particle size [56] becomes small (needle type). This can lead to a decrease in δ. Figure 7c shows that on an average τc increases with increase in temperature. The cloud becomes more opaque at higher temperatures. This has important implication in the radiative effects of STCs [38] in the context of global change.

## **7. Mean annual variation of particulate scattering in the UTLS region**

Figure 8a shows a contour plot of the logarithm of mean βp as a function of month and altitude for the period 1998–2003. The values of βp are relatively large during the May to September period and small during the winter months. This is due to the influence of STC. Relatively high values of βp in the UT region during the monsoon period are due to the presence of relatively dense cirrus and low values during winter are due to the presence of SVC. In contrast to the UT region, βp in the LS is generally large (as high as 6 × 10-8 m-1 sr-1) during the winter (November to January) and pre-monsoon (April–May) months and low (as low as 10-9 m-1 sr-1) during the summer (July and August) months. Prominent peaks are observed during May–June and November–January periods with low values in July–August and February. The mean annual pattern of αp at different altitudes from the lidar derived β<sup>p</sup> (Figure8a) is presented in Figure 8b. Similar grading scheme is used in both these plots to make an easy direct visual comparison of the pattern. The major features of the annual variation of βp and αp in different altitudes are very similar in these two plots, even though the lidar ratio is assumed to be variable with altitude depending on aerosol properties.

Microphysical properties like the size and shape of particles in the UTLS region can be delineated from the depolarization of backscattered radiation [57,58]. Figure 8c shows a contour plot of monthly mean δ in the altitude region 8–28 km. The value of δ varies in the range 0.03 to 0.6 in UT region and from 0.03– 0.04 in the LS region. High values of δ are

generally confined to a narrow altitude region (14–16 km) during winter while these extend to a wider altitude region (12– 18 km) during summer monsoon period. High values of δ (>0.2) are observed in the altitude region 14– 16 km during April–October period. Values of δ exceeding 0.04 observed in the UT (above 10 km) are mainly due to presence of highly non-spherical ice particles associated with STC [37]. The overall low values of δ suggests that particles in the LS region are very small and tend to become more spherical in nature.

Distribution of Particulates in the Tropical UTLS over

the Asian Summer Monsoon Region and Its Association with Atmospheric Dynamics 125

**Figure 9.** Contour plots of mean particulate extinction coefficient (αp), in logarithmic scale, at 525nm and 1020 nm as a function of month and altitude for the period 1998–2003 from SAGE-II observations

**Figure 10.** (a) Mean annual variation of cold point tropopause temperature and altitude along with the

in the range 16–18 km (a transition region), such that the region defined as UT is always below cold point and region defined as LS is above. Figure 10a shows the mean annual

The annual variation of δ for different altitudes in UT is shown in Figure 10b and that in LS in Figure 10c. Though, in general, on an average, the value of δ in LS region is < 0.04, it

mean annual variation of integrated volume depolarization ratio (VDR) normalized to the slab thickness of 2 km at different altitudes (b) in the upper troposphere, (c) in the lower stratosphere and

variation of the cold point tropopause altitude and tropopause temperature.

(d) in the tropopause region.

**Figure 8.** Contour plot of mean particulate backscatter coefficient (βp) and extinction coefficient (αp) in logarithmic scale along with the volume depolarization ratio (δ) as a function of month and altitude for the period 1998–2003 derived from lidar data. Mean cold point tropopause altitude is superposed over the contour along with its standard deviation.

Figure 9 shows a contour plot of αp similar to Figure 8b but generated from SAGE-II derived mean particulate extinction at 525 nm and 1020 nm over a small geographical grid size of 10- 16°N and 73-86°E centered around Gadanki. As thin STCs in the UT region significantly attenuates the SAGE-II wavelengths, (especially that at 525 nm) there will be a large data gap at the lower altitudes. The region bound between X-axis and the rectangular vertical bars (shown white) are the data gaps. As the attenuation for 1020 nm is less than that for 525 nm this wavelength can penetrate to lower altitudes to yield useful data. The data gap is relatively less for 1020 nm. However, in generating the contours, the data gap is appropriately interpolated. General similarity of the pattern in Figure 9a and 9b suggests that the interpolation did not influence the major features of Figure 9b. Except for an overall decrease in the values of αp derived from SAGE-II (at 525nm) compared to those derived from lidar data, the major spatio-temporal features in Figure 9 also matches well with those of Figure 8b. Thus, the inferences derived from lidar data is reconfirmed by SAGE-II observations during the same period.

To make the features more concise, the month to month variation of altitude weighted δ (altitude integral of δ normalized to the slab thickness) for different altitude regions with a slab thickness of 2 km are examined in the UT and LS region. Though a sharp definition of UT and LS region is rather difficult, for the present analysis we use the term UT for the altitude region from 10 to 16 km and LS from 18 to ∼30 km. Though the cold point tropopause shows a small variation with time of the day and day of the year, it lies always

the contour along with its standard deviation.

observations during the same period.

generally confined to a narrow altitude region (14–16 km) during winter while these extend to a wider altitude region (12– 18 km) during summer monsoon period. High values of δ (>0.2) are observed in the altitude region 14– 16 km during April–October period. Values of δ exceeding 0.04 observed in the UT (above 10 km) are mainly due to presence of highly non-spherical ice particles associated with STC [37]. The overall low values of δ suggests that particles in the LS region are very small and tend to become more spherical in nature.

**Figure 8.** Contour plot of mean particulate backscatter coefficient (βp) and extinction coefficient (αp) in logarithmic scale along with the volume depolarization ratio (δ) as a function of month and altitude for the period 1998–2003 derived from lidar data. Mean cold point tropopause altitude is superposed over

Figure 9 shows a contour plot of αp similar to Figure 8b but generated from SAGE-II derived mean particulate extinction at 525 nm and 1020 nm over a small geographical grid size of 10- 16°N and 73-86°E centered around Gadanki. As thin STCs in the UT region significantly attenuates the SAGE-II wavelengths, (especially that at 525 nm) there will be a large data gap at the lower altitudes. The region bound between X-axis and the rectangular vertical bars (shown white) are the data gaps. As the attenuation for 1020 nm is less than that for 525 nm this wavelength can penetrate to lower altitudes to yield useful data. The data gap is relatively less for 1020 nm. However, in generating the contours, the data gap is appropriately interpolated. General similarity of the pattern in Figure 9a and 9b suggests that the interpolation did not influence the major features of Figure 9b. Except for an overall decrease in the values of αp derived from SAGE-II (at 525nm) compared to those derived from lidar data, the major spatio-temporal features in Figure 9 also matches well with those of Figure 8b. Thus, the inferences derived from lidar data is reconfirmed by SAGE-II

To make the features more concise, the month to month variation of altitude weighted δ (altitude integral of δ normalized to the slab thickness) for different altitude regions with a slab thickness of 2 km are examined in the UT and LS region. Though a sharp definition of UT and LS region is rather difficult, for the present analysis we use the term UT for the altitude region from 10 to 16 km and LS from 18 to ∼30 km. Though the cold point tropopause shows a small variation with time of the day and day of the year, it lies always

**Figure 9.** Contour plots of mean particulate extinction coefficient (αp), in logarithmic scale, at 525nm and 1020 nm as a function of month and altitude for the period 1998–2003 from SAGE-II observations

**Figure 10.** (a) Mean annual variation of cold point tropopause temperature and altitude along with the mean annual variation of integrated volume depolarization ratio (VDR) normalized to the slab thickness of 2 km at different altitudes (b) in the upper troposphere, (c) in the lower stratosphere and (d) in the tropopause region.

in the range 16–18 km (a transition region), such that the region defined as UT is always below cold point and region defined as LS is above. Figure 10a shows the mean annual variation of the cold point tropopause altitude and tropopause temperature.

The annual variation of δ for different altitudes in UT is shown in Figure 10b and that in LS in Figure 10c. Though, in general, on an average, the value of δ in LS region is < 0.04, it

shows a pronounced oscillation with prominent peaks in May and September–January period. Above 20 km, the value of mean δ (~0.03) is very close to the molecular depolarization, it is relatively large (in the range 0.03 – 0.04) in the altitude region 18–20 km just above the cold point tropical tropopause. Thus, in the LS region, the particle size and non-sphericity decreases with increase in altitude. Note that, the value of δ encountered in this region is less than the threshold value (0.04) used for identifying the cirrus particles. The value of δ is largest in the altitude region 14–16 km, just below the mean level of tropical tropopause. The general similarity in the annual variation of δ in the altitude regions 14–16 km and 10-12 km indicates that the particle microphysics in these altitudes are strongly coupled. Even though the cold point tropopause altitude shows a small variation from month-to-month, the mean level lies around 16.5 km. Figure 10d shows the mean annual variation of the altitude weighted δ around the cold point tropopause for a slab thickness of 2 km. The observed general similarity of the annual variation of δ in the entire region from 10 to 20 km indicates that the annual variation of the particle habit (size and shape) in the UT and LS regions are strongly coupled. The observed general decrease in δ from UT to LS suggests that the particles tend to become small and more regular in shape with increase in altitude.

Distribution of Particulates in the Tropical UTLS over

the Asian Summer Monsoon Region and Its Association with Atmospheric Dynamics 127

Meteorology Department (Web site http://www.weather.uwyo.edu/upperair/ sounding.html.) are presented in Figure 11b. The CAPE also shows a prominent peak during the period April to May along with a small secondary peak in September. Intense convection is closely related to thunderstorm activity. Climatologically averaged thunderstorm activity in different latitude belts over the Indian subcontinent was investigated in detail in earlier study [63]. The annual variation of the mean number of thunderstorm days for the latitude belt 10–15°N (Figure 11c) shows two prominent peaks in May and September–October period. Deep convection and high thunderstorm activity leads to the formation of thick convective clouds. The main reason for low lidar observation statistics during the monsoon period is the presence of these convective clouds which impede the observations. Though these clouds will have a large spatial extent and persist for several days, some gap region which is devoid of thick clouds (favouring lidar observation) starts developing after a few days The outflow from adjacent convective anvils spreading

**Figure 11.** Annual variation of the (a) frequency distribution of outgoing long-wave radiation (<200 Wm-2) around Gadanki between 12.5°–15°N and 77.5°–80°E along with (b) convective available potential energy (CAPE) obtained from Radiosonde observations at Chennai during the period 1998– 2003. (c) Month-to-month variation of mean number of thunderstorm days (TSD) for the latitude belt

Being originated from convective outflow [43,44], especially during the summer monsoon period, particles of STCs in the UT region during this period will be relatively large and highly non-spherical [8,45]. Presence of these large non-spherical particles leads to an increase of δ in this region, as is observed. Subsequent uplift of some of these particles along with tropospheric air across the tropopause leads to an increase in δ along with the integrated backscatter (ΙβP) in the region just above the cold point tropopause. As the particles and precursor gases are not directly injected into the stratosphere but diffuse

over to this gap region leads to the formation of STCs.

10–15°N during the period 1970–1980 [63].
