Climatology of Stratospheric Ozone Based on SBUV and SBUV/2 Data:

Transcription

1 - -- NCAR/TN-42+STR NCAR TCHNCAL NOT L a April 995 Climatology of Stratospheric Ozone Based on SBUV and SBUV/2 Data: William J. Randel Fei Wu 8 N 6 OCT MAN COLUMN OZON (DU) S W ATMOSPHRC CHMSTRY DVSON NATONAL CNTR FOR ATMOSPHRC RSARCH BOULDR, COLORADO

2 Preface This atlas presents climatological stratospheric ozone statistics derived from nearly sixteen years ( ) of daily global satellite observations. Data from Nimbus 7 Solar Backscatter Ultraviolet (SBUV) (covering November 978-June 99) and NOAA SBUV/2 (January 989-April 994) are combined into a continuous time series; both column ozone and ozone profile information over -5 km is presented. The SBUV data are adjusted to match the SBUV/2 measurements, based on comparisons during the overlap period when both instruments were operational (January 989-June 99). The quality of this long-term ozone record is studied by extensive comparisons with other satellite ozone data, namely TOMS data for column ozone and LMS, SAG, HALO and MLS for ozone profiles (acronyms are identified below). The long term record is used to document climatological means, along with daily and interannual variability statistics, for zonal mean and planetary wave variations in ozone column and profile data. ly mean cross sections are presented, along with climatological latitude-time and height-time sections. This atlas is presented for a concise climatology of stratospheric ozone, and to alert potential users to the availability of these data.

3 Acknowledgments This work was completed in the Atmospheric Chemistry Division at NCAR, with support from NOAA grant NAAZ49 and NASA grants W-6 and W-88. The SBUV data were obtained from the National Space Science Data Center, and the SBUV/2 data were purchased from NOAA. We thank the TOMS group at NASA Goddard for access to Nimbus 7 and Meteor TOMS data, and Chip Trepte of NASA Langley for providing SAG data. LMS, HALO and MLS ozone data were obtained from the UARS group at NCAR under John Gille. We thank Jim Russell of NASA Langley and Joe Waters of JPL for access to and discussions regarding the HALO and MLS data, respectively. Dan Packman mapped all the SBUV and SBUV/2 data analyzed here. We thank Jim Gleason of NASA Goddard for insightful discussions on the data quality, and Rich McPeters of NASA and Rolando Garcia of NCAR for reviews of this manuscript. Marilena Stone expertly prepared the text. NCAR is sponsored by the National Science Foundation.

4 Table of Contents. ntroduction Data Description... 2 a. SBUV data... 2 b. SBUV/2 data c. Data mapping Adjustment of SBUV Data and Comparisons With Other Data Sets... 5 a. Adjustment of SBUV data Colum n ozone Ozone profiles b. Comparison with other ozone measurements Colum n ozone Ozone profiles Summary of profile comparisons SBUV-SBUV/2 Climatology a. Statistical calculations b. Omission of spurious data during volcanic aerosol events References Results a. ly average meridional cross sections b. -longitude maps c. nsemble latitude-time diagrams d. nsemble height-time diagrams e. Time series of monthly means f. Tables of monthly zonal means

5 . ntroduction Monitoring of stratospheric ozone from satellites provides near-global observations on a daily basis, allowing study of its mean structure and variability over a range of space and time scales. Such observations are valuable for detecting anthropogenically-caused ozone depletion, and also for evaluating model simulations of stratospheric ozone. For both of these uses, it is important to understand both the mean or average structure of stratospheric ozone (and its seasonal cycle), and also its 'natural' variability (that associated with 'normal' meteorological fluctuations). The purpose of this study is to construct and analyze a long-term, continuous record of satellite ozone data to document such means and variabilities. The data used here are a combined time series from the Solar Backscatter Ultraviolet (SBUV) instrument on the Nimbus 7 satellite (covering November 978-June 99) and the SBUV/2 instrument on NOAA (January 989-April 994). Retrievals from these instruments provide estimates of the ozone column and vertical profile of ozone over approximately -5 km (with vertical resolution near km). Details of both the SBUV and SBUV/2 data sets are described in Section 2, along with a discussion of the mapping used here. n order to combine the two data sets into one continuous record, measurements from the overlap period (January 989-June 99) are analyzed. Comparisons and adjustments are calculated as a function of solar zenith angle (sza) of the satellite measurements, because ) the original retrieval is dependent on sza, and 2) we find that the differences are often a compact, monotonic function of sza. Column ozone differences between SBUV and SBUV/2 are small (-.5%), except at very high sza (- 5%). For ozone profiles, larger SBUV-SBUV/2 differences are found (- %), and these differences depend on sza, altitude and hemisphere. Using these derived differences, the entire SBUV record over is adjusted to be consistent with the SBUV/2 data, and the SBUV/2

6 data is then appended beginning in January 989. Details of these adjustments are described in Section. The quality of these combined and adjusted data are compared to other satellite ozone measurements in Section. Column ozone is compared with version 6 (V6) Nimbus 7 TOMS data, and profile data is compared with LMS, SAG, HALO and MLS measurements (acronyms are defined in Section ). Biases of monthly and zonal mean data are of order 2-5%, with standard differences of 5-%. Section 4 contains a climatological atlas of means and variabilities of stratospheric ozone based on these combined-adjusted SBUV-SBUV/2 data. We analyze monthly means, and daily and interannual variability for zonal mean ozone, along with various stationary and transient planetary wave statistics. -time and height-time sections of means and variabilities are presented, together with Tables of monthly, zonal means for handy reference. 2. Data Description a. SBUV data The Nimbus 7 SBUV data analyzed here are the version 6 (V6) column ozone and ozone profile retrievals for the period November 978 to 2 June 99. Details of the instrument and data retrieval are discussed in Bhartia et al. (984), Fleig et al. (99), and McPeters et al. (994) (and references therein); only a brief description is included here. The Nimbus 7 satellite operated in a sun-synchronous orbit, crossing the equator near noon local time. SBUV is a nadir viewing instrument, making measurements along the orbit track from 8 N to 8 S, with orbits spaced approximately 26' apart in longitude. SBUV measures sunlight backscattered from the earth and atmosphere in 2 discrete wavelength bands, and also measures the solar irradiance; an ozone retrieval is performed using the ratio of backscatter to incident flux. Column ozone is derived from the four longest wavelength bands based on a table look up and interpolation process, in which measured radiances are compared with theoretical values (Klenk et al., 982; Fleig et al.,

7 99). The pre-computed table contains backscattered radiance as a function of total ozone and measurement geometry, and is computed based on a set of climatological ozone and temperature profiles. The SBUV ozone profile is derived in layer amounts using an optional statistical estimation algorithm (Mateer, 977; Rodgers, 976, 99). An a priori climatological ozone structure and its covariance patterns versus height are used in the estimation scheme, and for V6 retrievals this climatology is derived from ozonesonde and SAG satellite data. The optimum ozone profile solution is determined by iteratively perturbing the a priori profile until the radiances and integrated column ozone agree with the measured radiances and previously derived column ozone to within a prescribed uncertainty. Although the primary retrieved quantity is ozone layer amounts (in layers approximately 5 km thick), the retrieval algorithm also computes ozone mixing ratios at individual pressure levels by fitting a cubic spline to the cumulative ozone amounts. For the results here, we analyze mixing ratios (in parts per million by volume, ppmv) at pressure levels:, 2,, 7, 5,, 2,,.5 and. mb. The vertical resolution of the SBUV data is - 8 km above the peak of the ozone layer (- mb); near and below the ozone peak the resolution degrades to - 5 km (Rodgers, 99; McPeters et al., 994). Profile information below the peak is mostly constrained by the column ozone amount and a priori profile shape; there is no real information from SBUV in the lower stratosphere. The estimated errors of individual SBUV profiles is 5-% below 5 km, and 5% near 45 km. The error increases rapidly above 5 km (Bhartia et al., 984; Cunnold et al., 984). Several aspects of data quality are worth mentioning. First, before 98 the SBUV was turned off one day in four because of power limitations, resulting in regularly spaced data gaps. Second, the diffuser plate that was used to measure solar irradiance degraded with time, causing apparent trends in the ratioed radiances. The V6 SBUV data includes a correction for this effect (Herman et al., 99). Finally, beginning in February 987 the SBUV data were affected by a chopper wheel synchronization problem, which introduced significant noise into the data. Gleason et al. (994) devised a correction scheme for this

8 problem; although data during this period are considered of lower quality, there are no apparent systematic biases. After June 99 the synchronization problem became larger, and the data are not correctable. b. SBUV/2 data The NOAA SBUV/2 data analyzed here are column ozone and ozone profile retrievals using the Version 6 (V6) algorithm (almost identical to the SBUV algorithm). Data are analyzed covering the period January 989 to April 994. The SBUV/2 spectrometer is an improved version of the SBUV instrument, incorporating an on-board calibration lamp, and a change in one of the wavelengths observed (Planet et al., 994; Ahmad et al., 994). The absolute calibration of SBUV/2 was determined using space shuttle SBUV (SSBUV) data, as discussed in Planet et al. (994) and Hilsenrath et al. (994); similar adjustments were not made for the Nimbus 7 SBUV data, and this is one source of the differences between the two. The wavelength change included in SBUV/2 (replacing the 6-nm SBUV channel with a 2-nm channel) is meant to improve the ozone profile retrieval in the.7-.5 mb region. The on-based calibration for SBUV/2 measures in-orbit changes of the solar diffuser plate reflectivity over time, and is verified by comparisons with SSBUV measurements. The post-launch calibration has been extrapolated since January 99 based on the assumption that the rate of diffuser degradation has not changed. The NOAA orbit is nearly sun synchronous, but the equator crossing time drifted from its initial : pm to approximately 4: pm (in 994). This orbit drift results in the terminator being crossed at successively lower latitudes in winter months, limiting the latitudinal coverage (no data is available in darkness). This increasing loss of data with time is seen in some of the cross sections below.

9 c. Data mapping The SBUV and SBUV/2 orbital satellite data have been objectively analyzed or mapped here using a Kalman filter, as described by Rodgers (977). This produces an estimate of the zonal mean and the first six zonal harmonic wave coefficients on a 4-degree latitude grid from 8 S to 8 N. The Kalman filter is applied to the data in both forward and reverse time, and the results are averaged and weighted by the inverse of their errors to give an optimal estimate, which is sampled once per day.. Adjustment of SBUV Data and Comparisons With Other Data Sets a. Adjustment of SBUV data n order to combine the SBUV and SBUV/2 data sets into a continuous time series, we make comparisons for the overlap oeriod time (January 989-June 99), and use these differences to adjust the SBUV data. The choice of adjusting either the SBUV or SBUV/2 data is somewhat arbitrary, and we choose to adjust the SBUV based on ) the improved calibration of SBUV/2, and 2) our comparisons with other satellite ozone measurements below (overall smaller differences are found by adjusting SBUV rather than SBUV/2). For simpliicity, only monthly means are considered when comparing SBUV and SBUV/2 here. Differences are calculated and adjustments are made as a function of solar zenith angle of the SBUV measurements, because ) the original ozone retrievals are made as a function of solar zenith angle, and 2) we find compact relationships between the SBUV-SBUV/2 differences and the SBUV solar zenith angle. Differences and adjustments are also calculated separately for each hemisphere, because systematic differences are observed between NH and SH statistics (these differences are in turn related to the a priori information input into the retrieval scheme, and the different structure of ozone in the respective hemispheres). When adjustments are made to the SBUV data, only the zonal mean is changed; zonal wave components (which are typically of order % of zonal mean values) are left unchanged. We consider column ozone and profile ozone in turn. 5

10 . Column ozone Figure shows latitude-time sections of SBUV and SBUV/2 column ozone for the overlap period, along with their percentage difference. Differences less than 2 percent are seen over broad regions of latitude, except for high latitude winter measurements. Here differences of order 5% are typical, with positive differences in the NH (SBUV/2 > SBUV) and negative differences in the SH. Figure 2 shows the differences plotted versus solar zenith angle (sza), showing small differences (< 2%) for sza < 7, and systematic increase at higher values. The hemisphere mean SBUV/2-SBUV difference (area weighted average of the points in Fig. 2) and standard deviation (the spread of the differences about this mean) are.6 ±.5 DU for the NH and -.2 ± 2. DU for the SH. The adjustments applied to the column ozone are simply the mean percentage differences as function of sza, derived by averaging the individual points in Fig. 2 (as indicated by the line drawn through the points). Figure shows time series of column ozone at 64 N and 64 S for the entire record , comparing original and corrected SBUV data. Note the slight change in seasonal cycle in the adjusted data, and oppositely-signed changes in either hemisphere. 2. Ozone profiles Figure 4 shows the zonal mean ozone mixing ratio for January 989 from SBUV, SBUV/2 and their difference. SBUV/2 is slightly larger than SBUV (- 2%) below the profile maximum (-2 mb), while SBUV is larger in the upper stratosphere (- 8% difference near 2 mb). The differences are not uniform with latitude, and change between NH and SH. Figure 5 shows the SBUV/2-SBUV differences sampled over the entire 8 month overlap record, plotted as a function of sza, and separated by hemisphere. The thin line through the center of the points shows the average difference versus sza, and these are the actual corrections applied to the SBUV data. Note Fig. 5 only shows results at the,, and mb levels, but similar calculations and corrections are applied at each of the ten pressure levels.

11 Some notable features of the differences in Fig. 5: ) The differences are reasonably compact functions of SBUV solar zenith angle, and systematic changes are observed at high sza. 2) Differences are largest near and 2- mb (see Fig. 6 below), with a change in sign between these levels. These oppositely-signed differences will partly compensate when column ozone is calculated. ) There are clear NH-SH differences in the sza dependences, particularly at and mb. Figure 6 shows the hemisphere average differences and standard deviations for the SBUV/2-SBUV profile statistics. The average differences (or biases) are of order -5%, with one sigma variations near %. The differences are largest near 2- mb, mb, and mb, with the differences being of opposite sign between these levels. Figure 7 compares time series of original and adjusted SBUV mixing ratios at mb, and latitudes 52 N,, and 52 S. At the equator the adjustments are nearly constant throughout the record, because these are always low sza measurements. At 52 S the corrections are also nearly constant in time, due to the small sza dependence of the SH differences at mb in Fig. 5. Conversely, the adjustments at 52'N substantially affect the seasonal cycle amplitude, because of the strong sza dependence of the NH mb statistics in Fig. 5. 7

12 SBUV OZON ZONAL MAN (DU) e% Li UN 6N SON Q) N - -J S one i i i i, i _. VW ) Year.UN oskl SBUV2 OZON ', -. ZONAL MAN (DU),,.-, N SON N -a -,J S nn A VU5 -_f- i i Year.. t r i... SBUV2-SBUV ZONAL ^f.l MAN DFF (%) Q)._ --J Year Figure. -time plots of column ozone (DU) during January 989-June 99 from SBUV (top) and SBUV/2 (middle). Bottom panel shows the percentage difference, (SBUV/2-SBUV)/SBUV/2, with contour interval of 2%.

13 2 L, L- o LiJ z - N SBUV2-SBUV north.±. -,', ~.i.... +,'.-.p-,-, ' SOLAR ZNTH ANGL 9 X 2 L, Lo c z - N SBUV2-SBUV south SOLAR ZNTH ANGL + + Figure 2. Percentage differences in column ozone between SBUV and SBUV/2, plotted as function of solar zenith angle of the SBUV measurements. ach point represents the difference for one month at one latitude. The thin solid line denotes the mean difference at each solar zenith angle, and this is the adjustment applied to the SBUV data. The numbers in the upper left hand corers show the mean difference and standard deviation averaged over all solar zenith angles. 9

14 Column ozone 64N Column ozone 64S D Figure. Column ozone at 64 N (top) and 64 S (bottom) for the entire SBUV-SBUV/2 record. Thin dashed lines show the unadjusted SBUV data, while the solid lines (over 978-9) show the adjusted data. Data voids at the end of the record are due to absence of SBUV/2 data, due to the drift of the NOAA orbit.

15 .,,,,... Jan 989 SBUV OZON (PPMV) 5 45 " V /^^ ^~ i i:::::::: f(a, '4 ) 5 8S 4S 2S 2N 4N 6N 8N, $, a 2.~~~~~~~~~~iiiiiii. -,,.,..Jan 989 SBUV2 OZON (PPMV) iiiiii -ṟ i:: i iiii C?, o.,- i '_, i --- i. i. i. i. i. i 8S 4S 2S 2N 4N 6N 8N Jan 989 (SBUV2-SBUV)/SBUV2 (%) ) ,.... *. p P P P pp' ~:ii.iii Pt p P p p pp ( /.. : - P ~ p r, ~ ~~ *:::$pp.~~~~~~~~~~~p.-, Pp /'~~~~~::iiii. ' 8S 4S 2S 2N 4N 6N 8N Figure 4. Meridional cross sections of zonal mean ozone (ppmv) for January 989 from SBUV (top) and SBUV/2 (middle). Bottom panel shows their percentage differences.

16 2 L LJ z - N -2 - L Li 2 - N -2 - L- L, LL z N. 2 LLJ Lz - N -2 - NH SBUV2-SBUV mb iu zl SOLAR ZNTH ANGL SBUV2-SBUV mb SOLAR ZNTH ANGL SBUV2-SBUV mb SOLAR ZNTH ANGL SBUV2-SBUV mb -2.9±+. o S_ LL LJ.z N N LJ OZ - N LLf 44 * ~~ *~4 _ < rf <? ^* ^^^^ ^^^ L- *~~~~~~~~~~~~~~~~~~~ SOLAR ZNTH ANGL 4, ,\ r u n 2 LL Li z - N -2 LJ z o N -JU O tn SH SBUV2-SBUV mb. i., 't " i -2.5± 2..,! * ' d i *-* ~~~ '*C + +" ~~~~+? i SOLAR ZNTH ANGL SBUV2-SBUV mb -4.7± 4. ; J^7~~~~~~~~~~~~~~~~~~~~~~~~~~~~. 4 * SOLAR ZNTH ANGL SBUV2-SBUV mb, i i. 2..e;~a~'' j ^^^^f , t , ^^^^ SOLAR ZNTH ANGL SBUV2-SBUV mb SOLAR ZNTH ANGL 4., -.5 Li.~~..... J! J -VJ i, Figure 5. SBUV/2-SBUV ozone mixing ratio percentage differences versus solar zenith angle. Results are shown for the,, and mb levels, separated according to NH (left) and SH (right) statistics. ach point represents an individual monthly mean at one latitude. The thin lines through the center of the data points shows the corrections applied to the SBUV data. Numbers in the upper left hand corners show the mean difference and standard deviation averaged over all solar zenith angles.

17 (SBUV2-SBUV)/SBUV2 north. 6 (SBUV2-SBUV)/SBUV2 south..c -U ' L, C c! - 5 -ac. 4 (D m Diff (%) Diff (%) Figure 6. SBUV/2-SBUV ozone mixing ratio percentage differences versus altitude for the NH (left) and SH (right). These statistics are derived from averages over all solar zenith angles. Horizontal bars at each pressure level denote plus/minus one standard deviation.

18 8 SBUV OZON 52N Omb.., r- : Q_ CL a A a a,,,,, i SBUV OZON Q mb,,..,,.,..-,,.,,..,,,. -- > r- 9 L O. A A A'- M i 9 8 SBUV OZON 52S mb ''* ' ' '.'i '.2.CL V'vVJn 5 a a a i f a a --, t Figure 7. Time series of mb ozone mixing ratio at 52 N (top), equator (middle), and 52 S (bottom), for the entire SBUV-SBUV/2 record. Thin dashed lines show the unadjusted SBUV data (over ), while the heavy line over this time period shows the adjusted SBUV data.

19 b. Comparison with other ozone measurements n this section brief comparisons between the combined-adjusted SBUV-SBUV/2 data and other ozone measurements are summarized, in order to give an indication of the overall quality of the climatology analyzed here.. Column ozone xtensive comparisons with ground-based Dobson spectrophotometer column ozone measurements have been made for both SBUV and SBUV/2 (W`MO, 994; Planet et al., 994). Figure 8 summarizes these comparisons over NH midlatitudes, showing the mean differences (biases) over the entire SBUV and SBUV/2 data record (this figure was provided courtesy of Rich McPeters of NASA). SBUV-Dobson differences are small (±.5-%) throughout SBUV/2-Dobson differences are somewhat larger (± -.5%), and show strong seasonality (SBUV/2 too large in winter and too small in summer). This seasonality is believed to be related to non-linearity in the SBUV/2 photomultiplier. We also compare the SBUV-SBUV/2 column ozone record here with Total Ozone Mapping Spectrometer (TOMS) V6 data from the Nimbus 7 satellite (covering November 978-April 99). These data have been widely used in analyses of column ozone trends and global variability (e.g., Stolarski et al., 99; Bowman, 989). The TOMS spectrometer is similar overall to SBUV, making measurements of backscattered and incoming solar radiation, and deriving column ozone by a table look-up procedure. TOMS and SBUV share the same diffuser plate used to measure the solar irradiance, and similar corrections have been applied to TOMS and SBUV V6 data to account for its degradation (Herman et al., 99). An important difference between TOMS and SBUV is that TOMS scans horizontally across the sub-orbit track; it has much higher spatial resolution than SBUV, and also samples across a range of solar zenith angles. For the comparisons here, we use TOMS data gridded by the NASA Goddard Ozone Processing Team, the so-called 5

20 GRDTOMS data (McPeters et al., 99). One note of importance here is that zonal means from GRDTOMS data sample and average over a wide range of solar zenith angle measurements (in contrast to the nadir viewing SBUV-SBUV/2 instruments). Figure 9 shows the adjusted SBUV minus TOMS differences versus sza, sampled over the beginning (978-8) and end (987-88) of the SBUV record. These plots show a near-constant offset of -.% (thought to be due to pre-launch wavelength calibration effects in TOMS), with strong increase in bias and increased variability at very high sza; the overall patterns are very similar between the early and late sample periods. (Note: Much of the scatter at high sza in Fig. 9 is due to the adjustments applied to the SBUV data (see Fig. 2); comparisons with unadjusted SBUV data are more compact, with the bias monotonically increasing with sza). Figure shows similar results comparing SBUV/2-TOMS, and here the results are separated into periods before and after the eruption of Mt. Pinatubo in June 99. Note the somewhat larger spread of differences, particularly at high sza, after Pinatubo. Problems with high sza retrievals for nadir-viewing instruments (such as SBUV and SBUV/2) for time periods of enhanced aerosol loadings (i.e., after volcanic eruptions) have been discussed by Bhartia et al., 99; they suggest that significant negative biases (- -%) can occur under such conditions. However, Figs. 9- show that such high sza biases occur throughout the SBUV-SBUV/2 record, and hence are not due to volcanic aerosol effects alone. The overall results in Figs. 9- suggest that the mean SBUV- SBUV/2 column ozone is biased low at winter high latitudes, but that the overall space-time variability (as documented below) will be reasonably similar to that derived from TOMS data.

21 DOBSON COMPARSONS LU LL LL U- LL C).C Ul Figure 8. Time series of average differences between SBUV-SBUV/2 and ground-based Dobson spectrophotometer column ozone data, averaged over locations spanning -6 N. Solid lines show SBUV differences, and dashed lines those for SBUV/2. These comparisons are provided courtesy of Rich McPeters of NASA. 7

22 LLz U LL- Nov 978-Dec _. 2. SBUV -TOMS r SOLAR ZNTH ANGL Jan O 2 L L-. o LJ z - O. N Dec 988 SBUV -TOMS SOLAR ZNTH ANGL 9 Figure 9. Adjusted SBUV-TOMS column ozone differences versus solar zenith angle, for statistics over (top) and (bottom). Largest biases at high sza are from statistics over the Southern Hemisphere (SH).

23 L- i, LLJ z N Jan 2 O May SBUV/2-TOMS J&! SOLAR ZNTH ANGL 9 before Pinatubo LL z O N Jun Apr 99 SBUV/2-TOMS ,, ~ ~,Y.+ -+ Z ~ + i.,,.,. i., SOLAR ZNTH ANGL after Pinatubo Figure. SBUV/2-TOMS column ozone differences versus solar zenith angle for statistics over January 989-May 99 (top--prior to the volcanic eruption of Mt. Pinatubo) and June 99-April 99 (bottom--after Pinatubo). 9

24 2. Ozone profiles Bhartia et al. (984) have reported extensive comparisons between SBUV profile data and rocket, balloon and ground-based Umkehr profiles; they report biases of less than % and standard differences of 8-5%. Here we compare the combined-adjusted SBUV- SBUV/2 record with several independent satellite measurements. Some of these comparisons repeat previously reported results (e.g., Cunnold et al., 984; McPeters et al., 994). We present the comparisons here based on data in-house at NCAR, and for a concise comparison of our adjusted data with other data currently used by the community. We also analyze mixing ratio data directly (as opposed to layer amounts), because that is the geophysical quantity of most interest here. One common feature of these other satellite measurements is that their vertical resolution is significantly higher than that of SBUV- SBUV/2. Only very brief descriptions of the other satellite data sets are included here; the reader is referred to the original instrument description references for further details. For simplicity, the comparisons here are calculated for monthly and zonally averaged data (the SBUV-SBUV/2, LMS and MLS are actual monthly means, whereas the SAG and HALO data sample over a succession of latitudes throughout a month, and are not true time means). Because there are large diurnal variations in ozone above mb, and these separate instruments measure at different local times, we only show results over the - mb levels. Limb nfrared Monitor of the Stratosphere (LMS) ozone data are available during October 978-May 979, covering the latitude range 6 S-8'N (Gille et al., 984). LMS derives ozone by measurements of emitted infrared radiation from the atmospheric limb. Figure shows the LMS-SBUV differences at,, and mb versus sza, together with summary statistics at all pressure levels. Mean biases are small (< %) except at -7 mb, where SBUV is larger than LMS by 5-%, at mb, where LMS is larger by - 2%. Standard differences are - 5%, except at mb ( %). There is a clear sza dependence in the biases (more negative with increasing sza) over pressure levels 7- mb.

25 Figures 2 shows comparisons between adjusted SBUV and Stratospheric Aerosol and Gas xperiment (SAG ) ozone measurements (McCormick et al., 989). SAG derives ozone by solar occultation measurements; we combine both sunrise and sunset SAG measurements here. SAG data are available from October 984 to the present, and a subset of data sampled over are shown in Fig. (other periods exhibit similar statistics). Mean biases are small (< 2%) over 2-2 mb, with standard differences of order 5%. At mb SAG is % higher than SBUV, similar to the LMS-SBUV bias (Fig. ). The differences do not show strong dependence on sza or hemisphere (although because of sampling considerations, sza is only sampled up to ~ 7 ). Figures shows statistics for SAG -SBUV/2, sampled over 989-May 99 (prior to Mt. Pinatubo). Overall the statistics are very similar to the SAG -SBUV comparisons, with small biases over 2-2 mb, and somewhat larger biases at and mb. The similarity with the SAG -SBUV statistics is not surprising given that the SBUV data were adjusted to match SBUV/2, but this comparison confirms the homogenity of the SBUV-SBUV/2 data record. There is not a strong sza dependence in the biases in Fig., except at mb. Figure 4 shows comparisons between SBUV/2 and Halogen Occultation xperiment (HALO) ozone measurements. HALO is a solar occultation instrument flown on the Upper Atmosphere Research Satellite (UARS) (Russell et al., 99), and comparisons here cover the period October 99-April 994, using HALO V6 retrieval data. SBUV/2 ozone data are biased 5% less than HALO over 7-.5 mb, with smaller differences (2-%) over - mb and at mb. Standard differences of 5-% are typical. Figure 4 shows systematic bias changes at high sza at and mb, with oppositelysigned changes between these levels. Figure 5 shows difference statistics between SBUV/2 and Microwave Limb Sounder (MLS) ozone measurements (Froidevaux et al., 994). MLS also flew on UARS, deriving ozone from microwave emissions observed at the atmospheric limb; comparisons here cover the period January 992-April 994, using MLS V retrievals. The comparisons 2

26 show SBUV/2 biased 5-% low throughout - mb, with standard differences of 4-8%. MLS-SBUV/2 differences exhibit sza dependences at and mb of opposite sense, very similar to the HALO-SBUV/2 differences in Fig. 4.. Summary of profile comparisons The overall biases between the SBUV-SBUV/2 data and other satellite ozone profile measurements (LMS, SAG, HALO and MLS) are rather small, of order 2-%, with the following exceptions: instrument pressure levels SBUV-SBUV/2 bias LMS -7 mb. + 7% mb -% SAG mb -6% mb - % HALO (V6) 7-.5 mb +5% MLS (V) - mb -5-% The overall biases are not consistent between the different comparisons; although large - % biases are seen at mb in LMS and SAG comparisons, this level does not appear as an outlier in the HALO and MLS differences. The lack of consistent biases suggests that the SBUV-SBUV/2 data do not suffer from systematic errors larger than 5-%; there is simply an inherent 5% difference in the independent satellite ozone profile measurements. Similary, although the biases in individual comparisons can be sza dependent, the sza dependence is different between the separate measurement techniques. Standard differences for the monthly, zonal means is typically 5-%.

27 4, v 2 LL, L.. a o L - z O-2 N LMS-SBUV mb * 4 + 4, 4. 4, +, ± ^ * ' SOLAR ZNTH ANGL 4 ; ' 2 LL a w - z O -2 N o ; 2 LzḶ LMS-SBUV mb 4* + W , + 4 * '~~~~ SOLAR ZNTH ANGL LMS-SBUV mb -6.8 J 4.6- %- CD =i 5 4 F LMS-SBUV,..... (A c m. - w - z -2 N \*,,, , SOLAR ZNTH ANGL 2.., -2-2 OZON DFF (%) 4 o v 2 LL Lh C o LJ - -2 N LMS-SBUV mb 2.6 ' [~.+ $+ +Sss,+4+. : ''^y^,.+, ~.,#~.++,+.+'+++ +., 4 4, 4..4, + _ 4~+~~* a * 4 * t * * i.*, *, t, SOLAR ZNTH ANGL Figure. LMS - adjusted SBUV ozone mixing ratio differences versus solar zenith angle, at,, and mb (left panels). Right hand panel shows summary statistics versus altitude, with statistics averaged over all solar zenith angles. Horizontal bars at each pressure level denote plus/minus one standard deviation. 2

28 SAG-SBUV mb ,, , + *+. t_ +, ' -2 f ' '' W. L-, SOLAR ZNTH ANGL SAG-SBUV mb [ SAG-SBUV n._ -. v $? -2-2, *.,44. A,. + t N _ SOLAR ZNTH ANGL SAG-SBUV mb : 4. ' 2 + Lo * O: _ Y_ T. L -: * OZON DFF (%) SOLAR ZNTH ANGL -o A P%- 4U. v 2 U- L, n O N - _ A rn SAG-SBUV mb ' ' * ' "' ' ' _'» ^, ~ j t - ** P * ,...., 44.. '' ' '. ± SOLAR ZNTH ANGL Figure 2. SAG - adjusted SBUV ozone mixing ratio differences versus solar zenith angle, at,, and mb (left panels), for data sampled during Right hand panel shows summary statistics versus altitude, with statistics averaged over all solar zenith angles. Horizontal bars at each pressure level denote plus/minus one standard deviation.

29 4 " 2 L.. o w - o -2 N o SAG-SBUV2 mb 4 * ** 4 4 * 4^f *h * ± 9.9 :+ ~ ~ ~.+ +, 4, SOLAR ZNTH ANGL 9 4 o 2 Ll L.. O * - O -2 N o - -Anr.W 4 ^ 2 L.. O L - -2 o SAG-SBUV2 mb ', '", *,!,,. i..a^^^e SOLAR ZNTH ANGL _* ^^ = SAG-SBUV2 mb ' 4 ^ SOLAR ZNTH ANGL r-.2 2 SAG-SBUV/ OZON DFF (%) 4, v 2 Li. L O L - O -2 N o - -An -w SAG-SBUV2 mb 44 ~ ~ 44~ ~ * 4. i44 i 4, i r,!44! e 96 * i~ur SOLAR ZNTH ANGL Figure. SAG - SBUV/2 ozone mixing ratio differences versus solar zenith angle, at,, and mb (left panels), for data sampled during Right hand panel shows summary statistics versus altitude, with statistics averaged over all solar zenith angles. Horizontal bars at each pressure level denote plus/minus one standard deviation.

30 6. 4, 2 HALO-SBUV2 -.4 A- 5.8 mb Li Z -2 O N -4 O * t 4* V :», SOLAR ZNTH ANGL 9 6 S 4 L 2 O Li Z -2 N -4 O HALO-SBUV2-4.8.t mb,.,.,., SOLAR ZNTH ANGL HALO-SBUV2 mb HALO-SBUV2 CD c C".L 2 h O LJi z -2 N OZON DFF (%) SOLAR ZNTH ANGL 6 4 HALO-SBUV2 2mb ''.'.... ' U 2 4 t4, + 4, +, #: LJ Z -2 N -4 -fn * +t + 4. L.. * ' SOLAR ZNTH ANGL Figure 4. HALO - SBUV/2 ozone mixing ratio differences versus solar zenith angle, at,, and mb (left panels). Right hand panel shows summary statistics versus altitude, with statistics averaged over all solar zenith angles. Horizontal bars at each pressure level denote plus/minus one standard deviation.

31 6 4 MLS-SBUV2 6.7 ± 5.4 lmb,_ 2 L z -2 N -4 O +, +,>, + 4 F e * + t * o 4 L 2 LU z -2 N C SOLAR ZNTH ANGL MLS-SBUV2 5.5 :.6 _ mb,4-44,4, o) SOLAR ZNTH ANGL dis' MLS-SBUV2 " C 6 4 MLS-SBUV2 mb -, i, 7.5 ± , cn cd cr LL 2 Li z -2 N -4 4 t+ *+.,44.+ C i? : 2 OZON DFF (%) -6!...i i SOLAR ZNTH ANGL 6 e 4 MLS-SBUV mb L ',+ e+ L, z -2 N -4-6 '.i..'. *.. ~~~~~d*+~~~~~~~~~~~ SOLAR ZNTH ANGL Figure 5. MLS - SBUV/2 ozone mixing ratio differences versus solar zenith angle, at,, and mb (left panels). Right hand panel shows summary statistics versus altitude, with statistics averaged over all solar zenith angles. Horizontal bars at each pressure level denote plus/minus one standard deviation. 27

32

33 4. SBUV-SBUV/2 Climatology a. Statistical calculations Statistics regarding zonal mean and planetary wave variations in column and profile ozone are calculated as follows (Vi represents the value of variable y for year i (i = -7 for ) and dayj (J= for January, etc.)): () ly means: (2) nsemble monthly means: J [w~i =- l; () j= N < ' >=- X[w]i (2) N i= where N is the available numer of monthly means over November 978-April 994. () Daily standard deviation (about the respective monthly means) -N J /2 Gdaily = - S J (Vi - M-) 2 () (4) nterannual standard deviation of monthly means (about the ensemble average) = <N m >2] (4) (int =- ([N], - <V (4) Note the daily and interannual variance estimates (2aiy and 2n) may be summed to give the total variance about the ensemble mean. When analyzing variations in zonal mean quantities, it should be noted that variance increases with latitude due solely to the spherical geometry of the earth. To alleviate this dependence here, all subsequent maps of zonal mean daily and interannual standard deviations are multiplied by the square root of the cosine of latitude (following North et al., 982). One point to note is that the interannual variability estimates calculated here include the effects of both 'natural' variability and 'trends.' At certain times and places, trends are 29

34 the dominant component (such as over high SH latitudes in October, associated with development of the Antartic ozone hole during ). No attempt is made here to separate trends from other components of variability. Ozone trends calculated from SBUV- SBUV/2 data are documented in WMO (994) and Hollandsworth et al. (995). Planetary wave statistics are calculated for total, stationary and transient rms components: N J totalrms = Y, v J+ CjJ (5) N i= J j/=k- k ik where VS/ik (VC k) is the zonal sine (cosine) coefficient for wave number k. Stationary waves are defined from the ensemble mean coefficients: RMS transient waves are then defined as: stationary rms = < S >k + < VC >k 2 (6) k=lj transient rms = [N (VS'>k) +VC - C>k)2) (7) N i=j j=k=l Note that (total rms) 2 = (stationary rms) 2 + (transient rms) 2. We also separate individual zonal wave number components by restricting the zonal wave number (k) summation in (5)- (7). -time and height-time sections are displayed based on the monthly average quantities calculated above. A running -2- time filter is applied prior to plotting to slightly smooth the resulting statistics. b. Omission of spurious data during volcanic aerosol events xamination of the entire SBUV-SBUV/2 time record shows that periods after the volcanic eruptions of l Chichon (April 982) and Mt. Pinatubo (June 99) exhibit anomalous low ozone values over the tropics in the 7- mb region; Fig. 6 shows an example at mb, 8 N.. These decreases are not caused by photochemical or dynamical

35 processes, but reflect measurement errors by SBUV-type instruments in the presence of large aerosol loadings (Fleig et al., 99) (the aerosol does not strongly affect column ozone retrievals at low sza (Bhartia et al., 99)). nclusion of data after these volcanic eruptions results in excessive interannual variance, and 'bulls-eyes' in various plots (Fig. 7 shows an example of the June zonal mean interannual standard deviation aoin using the full data record). To alleviate these obvious data problems, we have simply omitted one year following each volcanic eruption from the statistics calculated here (as indicated in Fig. 6). Figure 7 shows the resulting June zonal mean aint, showing removal of the spurious bullseye centered near 2 N, -2 mb.

36 SBUV OZON 8N mb ' 4 ' - >C C:L 9 8, 7 7 F T lchichon l~ Chichron Pinatubo Figure 6. Time series of ozone mixing ratio (ppmv) at 8 N, mb. Time periods following the volcanic eruptions of l Chichon and Mt. Pinatubo (denoted by the dashed lines) are associated with spurious low ozone values, related to retrieval errors when excessive volcanic aerosols are present. These time periods are omitted from the climatology here.

37 Jun interannual sd (ppmv) : ::::: -\ _ ' _ *C' C, ( Full record.. -, t - - m,, 8S 4S 2S 2N 4N 6N 8N Jun interannual sd (ppmv) '" 4.c ) a) 5 2 *Xi r\,. :::::.. \ \ \i 8S 4S 2S 2N 4N 6N 8N CD C) t c J O Volcano periods omitted Figure 7. nterannual standard deviation of June average ozone mixing ratio, with and without inclusion of volcanic time periods (see Fig. 6). Note the 'bulls-eye' near 2 N in the top panel, related to spurious volcanic aerosol induced retrieval errors.

38

39 5. References Ahmad, Z., et al., 994: Accuracy of total ozone retrieval from NOAA SBUV/2 measurements: impact of instrument performance, J. Geophys. Res., 99, 22, ,984. Bartia, P. K., K. F. Klenk, A. J. Fleig, C. G. Wellemeyer and D. Gordon, 984: ntercomparison of NMBUS-7 Solar Backscatter Ultraviolet ozone profiles with rocket, balloon and Umkehr profiles, J. Geophys. Res., 89, Bhartia, P. K., J. Herman, R. D. McPeters and. Torres, 984: ffects of Mt. Pinatubo aerosols on total ozone measurements from backscatter ultraviolet (BUV) experiments, J. Geophys, Res., 98, 8,547-8,4. Bowman, K. P., 989: Global patterns of the quasi-biennial oscillation in total ozone, J. Atmos. Sci., 46, Cunnold, D. M., M. C. Pitts and C. R. Trepte, 984: An intercomparison of SAG and SBUV ozone observations for March and April 979, J. Geophys. Res., 89, Fleig, A. J., et al., 99: Nimbus-7 Solar Backscatter Ultraviolet (SBUV) ozone products user's guide, NASA Rep. 24, 28 pp. Froidevaux, L., et al., 994: Global ozone observations from the UARS MLS: an overview of zonal mean results, J. Atmos. Sci., 2, Gille, J. C. and J. M. Russell, 984: The Limb nfrared Monitor of the Stratosphere: xperiment description, performance and results, J. Geophys. Res., 89, Gleason, J. F., R. D. McPeters and J. R. Herman, 994: Characterization and analysis of the Nimbus-7 SBUV instrument in the 'non-sync' period (February 987-present), Proceedings of the Quadrennial Symposium on Ozone, NASA Conf. Publ. 266, pp

40 Herman, J. R., et al., 99: A new self-calibration method applied to TOMS and SBUV backscattered ultraviolet data to determine long-term global ozone change, J. Geophys. Res., 96, Hilsenrath,., et al., 994: Calibration of the NOAA- SBUV/2 ozone data set from using in-flight calibration data and SSBUV, J. Geophys, Res., in press. Hollandsworth, S., R. McPeters, L. Flynn, W. Planet, A. J. Miller and S. Chandra, 995: Ozone trends deduced from combined Nimbus 7 SBUV and NOAA SBUV/2 data, Geophys. Res. Lett., in press. Klenk, K. F., P. K. Bhartia, A. J. Fleig, V. G. Kaveeshwar, R. D. McPeters and P. M. Smith, 982: Total ozone determinatin from the backscattered ultraviolet (BUV) experiment, J. Appl. Meteorol., 2, McCormick, M. P., J. M. Zawodny, R.. Veiga, J. C. Larsen and P. H. Wang, 989: An overview of SAG and ozone measurements, Planet. Space Sci., 7, Mateer, C. L., 977: xperience with the inversion of Nimbus-4 BUV measurements to retrieve the ozone profile, in nversion Methods in Atmospheric Remote Sounding, edited by A. Deepak, pp , Academic, New York. McPeters, R. D., T. Miles, L.. Flynn, C. G. Wellemeyer and J. M. Zawodny, 994: Comparison of SBUV and SAG ozone profiles: mplications for ozone trends, J. Geophys. Res., 99, 2,5-2,524. McPeters, R. D., et al., 99: Nimbus-7 Total Ozone Mapping Spectrometer (TOMS) data products users guide, NASA Rep. 2, 85 pp. North, G. R., F. J. Moeng, T. L. Bell and R. F. Calahan, 982: The latitude dependence of the variance of zonally averaged quantities, Mon. Wea. Rev.,, Planet, W. G., et al., 994: Northern hemisphere total ozone from determined with the NOAA- Solar Backscatter Ultraviolet (SBUV/2) instrument, Geophys. Res. Lett., 2, -28.

41 Rodgers, C. D., 976: Retrieval of atmospheric temperature and composition from remote measurements of thermal radiation. Rev. Geophys. Space Phys., 4, Rodgers, C. D., 977: Statistical principles of inversin theory, in nversion Methods in Atmospheric Remote Sounding, edited by A. Deepak, pp. 7-8, Academic Press, New York. Rodgers, C. D., 99: Characterization and error analysis of profiles retrieved from remote sounding measurements, J. Geophys. Res., 95, Russell, J. M., et al., 99: The Halogen occulation experiment, J. Geophys. Res., 98,,777-,797. Stolarski, R. S., P. Bloomfield, R. D. McPeters and J. R. Herman, 99: Total ozone trends deduced from Nimbus-7 TOMS data, Geophys. Res. Lett., 8, 5-8. WMO, 994: Scientific Assessment of Ozone Depletion: 994, World Meteorological Organization Global Ozone Research and Monitoring Project, Report No. 7. 7

42

43 6. Results a. ly average meridional cross sections For each month there are three pages of figures; individual plots are set in the sequence described in a set of templates that follow, with contour intervals noted. Stippled regions denote polar night where no data are available. 9

44 mean ozone (ppmv) "- " - i ' i total rms (ppmv) ~~~~~~ * ' 4. 5 r nsemble monthly mean Zonal mean ozone (q. 2) Contours: ppmv S 4S 2S 2N 4N 6N 8N - s L L U) z 4 'Q 5 F Total RMS k=-6 planetary wave amplitude (q. 5) Contours:. ppmv , 8S 4S 2S 2N 4N 6N 8N r L (/) (D. - daily sd (ppmv) stationary rms (ppmv) ) =- ) 5 Zonal mean ozone daily standard deviation (q. ) Contours:.5 ppmv D * 4 u mo 5 - Sn - r F Stationary RMS k=-6 planetary wave amplitude (q. 6) Contours:. ppmv (n () L.. 8S 4S 2S. 2N 4N 6N 8N S 4S 2S 2N 4N 6N 8N interannual sd (ppmv) transient rms (ppmv),,,,,....,,.*. *.....,.., " 4._ ) 5 Zonal mean ozone interannual standard deviation (q. 4) Contours:.5 ppmv Transient RMS k=-6 planetary wave amplitude (q. 7) Contours:. ppmv Q () - Q) L S 4S 2S 2N 4N 6N 8N 8S. -. 4S 2S 2N 4N 6N 8N

45 stationary k= (ppmv) "i *.r- *. *..., *.. stationary k=2 (ppmv) '" 4 ) 5 Stationary k=l amplitude (q. 6 with k=l) Contours:. ppmv, 5 o - 45 V45 %-. L 4 ( ) 5 L Stationary k=2 amplitude (q. 6 with k=2) Contours:. ppmv o L /) S 4S 2S 2N 4N 6N 8N ~ 8S 4S 2S 2N 4N 6N 8N. transient k= (ppmv) transient k=2 (ppmv) _ " 4 D 5 r Transient k=l amplitude (q. 7 with k=l) Contours:. ppmv U) ' " 4 V) la 5 L Transient k=2 amplitude (q. 7 with k=2) Contours:. ppmv n L U. C :... -,.! 8S 4S 2S 2N 4N 6N 8N i. i i _.i 8S 4S 2S 2N 4N 6N 8N transient k= (ppmv) transient k=4-6 (ppmv) ) 5 Transient k= amplitude (q. 7 with k=) Contours:. ppmv 5 45 Transient k=4-6 amplitude,- (,(q. 7 with k=4-6) " 4 w.) Contours:. ppmv (D 5-5. L- (, L a S 4S 2S 2N 4N 6N 8N 8S 4S 2 2N 4N 6N 8N 4

46 r Jan mean ozone (ppmv) V [ " 4 ) , 5 4o *fi ffi () ) U* a ) 5 L Jan "'x total rms (ppmv) ' '" '. <'. 8S 2N 4S 4N 2S 6N -D L L -i 8S 4S 2S 2N 4N 6N 8N... 8S 4S 2S 2N 4N 6N 8N Jan daily sd (ppmv)..... *.. *... Jan stationary rms (ppmv) , 5-45 a ' 4._? ) S 4S 2S 2N 4N 6N 8N L 4 L n )Z a, 5 L o- C)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 4... ]. 8S 4S 2S 2N 4N 6N 8N L : () n Jan interannual sd (ppmv) ~ ~ ~~~~~..... * Jan transient rms (ppmv). a, = -C Un. m. a- sn U) ) L a_ S 4S 2S 2N 4N 6N 8N

47 5 45 _ 4.) 5 -- Jan stationary k= (ppmv)! i,:::::::! is... :< (^* ^ A ///l ::::;:::: :::)l l * \,::::::: J \ ::::::: i i/ / i 45.. ;:::::: s \ l :::::::::... { \ l :::::::::... J.to' ll/@oo~~~~~... W r- //~~~~~~~~... } *}!! A T 'o~~~~~~~~~~~~~~litit!~~~~~~~~....~~~~~~~~~~~~~~~~~~~~~~~~~~~~... 8S 4S 2S 2N 4N 6N 8N.! J Jan stationary k=2 (ppmv) S 4S 2S 2N 4N 6N 8N ''''''' '""" """' "'""' iiiiiiiii Do """"i """"',, """" L. l a_.....,... """", i Jan transient k= (ppmv) ~. r.~ - ~. r- - _. Jan transient k=2 (ppmv),.,.,..,,7 :., 5 ) 45 ' 4 r- C) -% SC D C n m L 5 al C L () (/ l- 2., S 4S 2S 2N 4N 6N 8O 2 _ 8S 4S 2S 2N 4N 6N 8N Jan transient k.= (ppmv). Jan transient k=4-6 (ppmv) ' -' ' ' ' ',, i ' i i :::. 5.,, 45 - Ė - 4 ) C 5 -r rc, ::: ::: iii ::::::::: :::: ::::: iiii ::::: ' L- 4 ) ) * s 5 L :D L L -- ',, r, r, r. r,, r ' S 4S 2S 2N 4N 6N 8N 8S 4S 2S 2N 4N 6N 8N 4

48 Feb mean ozone (ppmv) Feb total rms (ppmv) C) 5.b b - 5~~~~~ 2-- ~ S 4S 2S 2N 4N 6N 8N 8 7 ^ - -+ C. L - i..! L!,!..!. a,! 8S 4S 2S 2N 4N 6N 8N. -o L U) U) L~ Feb daily sd (ppmv)....,., Feb stationary rms (ppmv) 'F ) 5 ' t 8S 4S 2S 2N 4N 6N 8N 45.v n L. a. : 4 O=- a ernḻ - L S 4S 2S 2N 4N 6N 8N n U). - - Feb interannual sd (ppmv) - Ī Feb transient rms (ppmv) r - 7 i - / \ D-^a U) ) L a. L U) L i 8S 4S S 2S 2N 4N 6N 8N S 4S 2S 2N 4N 6N 8N ii^n o

49 Feb stationary k= (ppmv) Feb stationary k=2 (ppmv) ' 4 o) 5 ^, 5 45 _ 5 Q- 5 D L V). 8S S-- - w l w, 4S 2S 2N 4N 6N 8N 8S. i w. ~ ffi 4S 2S 2N 4N 6N 8N Feb transient k=l (ppmv) Feb transient k=2 (ppmv).? -nt =~ Q) V) ). - -o " 4 ) =- - Lu) U),L a- 8S 4S 2S 2N 4N 6N 8N Feb transient k= (ppmv) * * ' ~ **. *.. Feb transient k=4-6 (ppmv), i. [,, _ #,, 5 /" f 45 D- - \..,, ; -ou, 8S 4S 2S 2N 4N 6N 8N 7 r, 4 n ) S , S 4S 2S 2N 4N 6N 8N - L a: 45

50 Mar mean ozone (ppmv) Mar total rms (ppmv), ' - i f ', -!.,,, ' 4 5 / D - on CL * -m i L () L l- * 5~~~6 8S 4S 2S 2N 4N 6N 8f B O.6 S.... *.. 8S 4S 2S 2N 4N 6N 8N i Mar daily sd (ppmv) Mar stationary rms (ppmv) r- '" - 4 ) S 4S 2S 2N 4N 6N 8N cn L ;V r- JD 45 L '' c- 4 U) ) 5 4~ Qa i.a..... OS 4S 2S 2N 4N 6N 8N J -o L /: U) U). L a-' Mar interannual sd (ppmv)... ~ Mar transient rms (ppmv) ) 4 - r 5 n s L - D. L c) U) L a S 4S 2S 2N 4N 6N 8N 8S 4S 2S 2N 4N 6N 8N