--Jack A. Kaye (jkaye@hq.nasa.gov), Manager, Atmospheric Chemistry Modeling and Analysis Program, Office of Earth Science, NASA Headquarters
| The Science Team Meeting for the Stratospheric Aerosol and Gas Experiment (SAGE II) instrument was held at Hampton University in Hampton, VA, on February 12-13. SAGE II, which has been flying for over 13 years aboard the Earth Radiation Budget Satellite (ERBS), measures many of the same parameters that will be measured by the SAGE III instrument that forms part of the Earth Observing System. Measurements of vertical profiles of ozone, aerosols, water vapor, and nitrogen dioxide, as well as cloud presence, are made by SAGE II using the technique of extinction of solar radiation at occultation. The first SAGE III instrument, which will also allow for the measurement of additional species, will have lunar occultation capability. It will be launched aboard a Russian Meteor-3M satellite in the summer of 1999. A summary of the meeting is given below. |
The meeting began with an introduction and greetings by the SAGE II Principal Investigator, M. Patrick McCormick, of the Center for Atmospheric Sciences at Hampton University (HU). He summarized current activity in the Physics Department at HU, especially several new hires in the area of atmospheric chemistry and physics. A summary of recent activities related to NASA's newly renamed Office of Earth Science and planned research announcements likely to be of interest to the SAGE investigator community was presented by the SAGE II Program Scientist, Jack Kaye, of NASA Headquarters, including the planned recompetition for the SAGE II Science Team in an announcement expected to be released this spring with proposals due in early summer.
Science presentations began with a detailed discussion by Joseph Zawodny of the NASA/Langley Research Center (LaRC) of the status of the SAGE II instrument and reprocessing activities. The instrument continues to operate normally. Data are currently available through the LaRC Distributed Active Archive Center (DAAC) for the period from launch through January, 1998, using the current version of the algorithm (version 5.931). A new version of the algorithm (5.96) has been developed for only the ozone and aerosol products; it has been used to prepare revised data sets for the ozone trends activity carried out for the Stratospheric Processes and their Role in Climate (SPARC) subgroup of the World Climate Research Programme. A newer version of the algorithm, ultimately designed for providing new data sets for all SAGE products (6.0), is currently under development, with the new data sets to be released to the science community late this summer. The main new aspect of V5.96 is the use of a new model for estimating the 600-nm aerosol extinction that is needed if one is to determine ozone extinction from the total extinction at 600 nm; use of this model has significantly improved the retrieved properties, especially during the times of high aerosol loading following the Mt. Pinatubo eruption. The V5.96 ozone and aerosol data sets do have some unexplained diurnal variations, however, at high altitudes. Some problems in nitrogen dioxide vertical profiles found with the earlier version of the SAGE algorithm have been resolved (especially through correction of time-dependent properties of one of the two channels used in the NO2 detection), but at present the NO2 data set is not recommended for use in scientific studies. The version 6.0 algorithm will include several improvements, including the incorporation of multi-wavelength refraction, addition of an oblate Earth model to the representation of refraction, and corrections to the "edge times" observed as the SAGE II instrument scans across the solar disk. The first end-to-end test of the new algorithm was completed in January, 1998. Additional work to be done includes an increase of vertical resolution from 1 km to 0.5 km, improvements of the ozone retrieval above 40 km, incorporation of new NO2 spectroscopy, and improved representation of spectral interference between ozone and water vapor. When the new algorithm is fully implemented and a data set prepared, users will be needed to carry out validation activities.
Philip Russell of the NASA Ames Research Center reviewed work by his team to develop an integrated picture of the evolution of the pre- and post-Pinatubo stratospheric aerosol. The recent work has focused on combining SAGE II extinction measurements with those from the Cryogenic Limb Array Etalon Spectrometer (CLAES) instrument on the Upper Atmosphere Research Satellite (UARS). Previous work had shown how particle sizes increased for about a year after the eruption and decreased very slowly thereafter. However, those results also showed that when effective radius values exceed ~ 0.4 µm, an upper bound on particle size cannot be retrieved from optical depth spectra unless the spectra extend to wavelengths greater than 1 µm. Therefore, a technique for deriving effective radius values from extinction in the SAGE II wavelength range was extended to include UARS/CLAES-measured extinction at 7.955 µm. The technique takes into account the error bars on measured extinction at each wavelength and reflects them as error bars on retrieved values of effective radius, surface area S, and volume V. Interesting features in the results include: (i) increases in S and V (but not effective radius) after the Ruiz and Kelut injections, (ii) increases in the S, V, and effective radius after Pinatubo, (iii) post-Pinatubo increases in S, V, and effective radius that are more rapid in the tropics than elsewhere, (iv) midlatitude post-Pinatubo increases in effective radius that lag increases in S and V, and (v) S and V returning to pre-Pinatubo values sooner than effective radius does. This task also contributed to publications on the life cycle of stratospheric aerosols, stratospheric aerosol effects in General Circulation Model (GCM) studies of climate change, and simplified approaches to estimating radiative effects of aerosols. Future work will include examining longitudinal dependencies and determining and correcting for effects of assumption of a unimodal, lognormal size distribution.
Geoffrey Kent of Science and Technology Corp. summarized work he has done determining tropospheric aerosol distributions from SAGE observations. This work, which relies on a method for separating aerosol and cloud extinction, has been applied to the 13 years of SAGE data except for an approximately 2-year period following the Mt. Pinatubo eruption (1991-1993). SAGE I data have also been studied in this way although they have less information because of the non-availability of aerosol extinction data at 0.5 µm. Besides extending the previous work to now cover the post-Pinatubo period, the most recent work includes a relaxation of the assumption that there is an exponential decay for the volcanic component of the aerosol; now the volcanic component of the upper tropospheric optical depth is assumed to be linearly dependent on the stratospheric aerosol overburden within the same season and latitude band. Latitudinal distributions of upper tropospheric aerosols were obtained, showing an order of magnitude increase between 70° S and N. Long-term trends in upper tropospheric aerosol have been studied using the SAGE I and SAGE II data. It is unlikely that any change greater than 1% per year has occured in the upper tropospheric aerosol optical depth of either hemisphere.
Pi-Huan Wang, also of Science and Technology Corp., briefly described work he has done on the derivation of information on tropospheric meridional circulation using SAGE aerosol and cloud data. This work has been described in a paper which has been accepted for publication in J. Geophys. Res. The circulation that is derived in this work differs from that coming from meteorological models. Wang also noted his SAGE-related work in other areas, including studies of eddy isentropic transport of ozone in the "middleworld" and studies of the climatology of global cloud occurrence as seen in SAGE.
The geographical distribution of upper tropospheric aerosols as observed by SAGE II was reviewed by Chip Trepte of the NASA Langley Research Center. Maps of aerosols were produced when there was sufficient sampling in a given latitude/longitude region. One particularly interesting result to come from this analysis was an apparent "plume" of mid-tropospheric aerosols (6.5 km altitude) produced from biomass burning in South America that nearly circled the globe in southern midlatitudes. Comparisons of SAGE-observed aerosols with data obtained during the Lidar in Space Technology Experiment (LITE) Space Shuttle mission in 1994 were also carried out.
Comparisons of aerosol observations made from SAGE and from in situ instruments aboard the ER-2 and DC-8 aircraft were presented by Darrell Baumgardner of the National Center for Atmospheric Research. Data from several different airborne campaigns were reviewed, including the Antarctic Southern Hemisphere Ozone Expedition (ASHOE), the Photochemistry of Ozone Loss in Arctic Region in Summer (POLARIS), and the Tropical Ozone Transport Experiment/Vortex Ozone Transport Experiment (TOTE/VOTE). During the ASHOE mission, there were several nice matchups between the in situ and SAGE measurements, and similar values for the effective radius of stratospheric aerosol particles were found. The in situ instruments have shown evidence for asphericity in some of the aerosol particles; the origins of this is not well understood.
Additional in situ - SAGE comparisons were reviewed by Chuck Wilson of the University of Denver, who has been focusing on non-volcanic aerosols in his comparisons. A major difficulty in making such comparisons is the effects of having only approximate matchups between the in situ and SAGE measurements, as it can be difficult to interpret the significance of differences for near-matchups. In some cases, the differences between the two measurements are larger than would be expected for coincidences based on the properties of the measurements. Work on the relationship between aerosol distributions and those of their precursors was also presented; results suggest that the rate at which sulfate appears (based on comparison with the age of air) is consistent with that at which carbonyl sulfide, a key precursor, disappears. The in situ results show that in "older" air, the concentration of large particles (~ 1 µm diameter) is much smaller than in younger air. These results suggest that SAGE may be able to provide additional insight on aerosol size distribution. Remaining work centers on increasing understanding of stratospheric sulfate mixing ratios and the role of large particles in the stratosphere.
Simulation of stratospheric aerosols using a combined two-dimensional atmospheric chemistry model with a detailed microphysical model was presented by Michael Mills of the University of Colorado. This model includes a fairly detailed representation of aerosol microphysics (condensation/evaporation, coagulation, transport, sedimentation, homogeneous nucleation) and carries 45 size bins for aerosols. Aerosol-related chemical species include carbonyl sulfide, sulfur dioxide, and sulfuric acid. Through comparison with SAGE observations of stratospheric aerosol abundance, it appears that carbonyl sulfide cannot provide for enough aerosol mass in the stratosphere, especially in the lower stratosphere. It therefore appears that tropospheric aerosol and sulfur dioxide account for a sizable fraction of aerosol mass in the lower stratosphere. These results were independent of assumptions made about the long-wavelength absorption of carbonyl sulfide (longward of 295 nm) where laboratory data are not available.
Simulations of the Mt. Pinatubo aerosol cloud using a three-dimensional model were presented by Howard Houben of the Bay Area Environmental Research Institute. The simulations suggest that the self heating of the aerosol cloud is very important to how it spreads; if assumed to be non-interacting, the clouds would have less impact. Model-measurement comparison is improved if slightly larger particles are used than had been previously assumed. The model results and existing observations also suggest that following the Mt. Pinatubo eruption there was not a significant increase in the total number density of stratospheric particlesthat most of the condensation was onto existing particles, which became larger. In some cases, sedimentation of these larger particles may actually lead to a "cleaning" of the stratosphere, which is a somewhat unexpected result. The ability to test models during the period most closely following the eruption is limited by the scarcity of data during this time period.
Two-dimensional aerosol simulations were also presented by Debra Weisenstein of Atmospheric and Environmental Research, Inc. Her model included a fairly complete representation of atmospheric gas phase sulfur chemistry (including carbonyl disulfide, dimethyl sulfide, carbonyl sulfide, methane sulfonic acid, sulfur dioxide, sulfur trioxide, and sulfuric acid), aerosol microphysics, and size distributions (40 size bins from 0.39 nm - 3.3 µm). A particular question addressed was whether or not tropospheric deep convection would impact stratospheric aerosol loading. Results suggested that this was the case, as carbonyl sulfide yielded aerosol loading approximately a factor of two lower than observations. By imposing elevated sulfur dioxide (40 pptv) in the tropics and subtropics (equatorward of 24°) in the middle and upper troposphere (4-12 km), much better agreement between calculated and observed stratospheric aerosol extinction was obtained. A calculation for the post-Mt. Pinatubo period was also carried out, in which 20 Mton of sulfur dioxide were put uniformly between 5° S and 14° N and 16-29 km. The calculated e-folding removal rate for sulfur dioxide (39 days) was fairly close to the observed value (33-35 days), and the calculated peak aerosol loading amount (32.7 Mton) and time (4 mo.) were fairly close to that observed. Comparisons of calculated aerosol amounts with those determined from SAGE for 2 km above the tropopause showed that the model-calculated aerosols were removed somewhat more rapidly than those in the atmosphere. This model is also being used to study the potential effects of a projected fleet of high speed supersonic civilian transport aircraft.
Comparisons of SAGE II and Halogen Occultation Experiment (HALOE) water vapor observations with those made from in situ platforms, notably balloons and the NASA ER-2, were summarized by Adrian Tuck of the NOAA Aeronomy Laboratory. Some work, already published, showed that the Southern Hemisphere lower stratosphere is significantly drier than that of the Northern Hemisphere for three major reasons the relationship between the annual cycle in tropical tropopause temperature and that of the Brewer-Dobson circulation, desiccation in the Antarctic in winter, and differences in high-latitude wintertime descent between the two hemispheres. Work currently underway will extend this analysis to look at longitudinal variability in water vapor concentrations, especially the greater desiccation observed at high southern latitudes near the Greenwich meridian relative to the international date line; this is true both inside and outside the vortex. Studies of long-term trends in stratospheric water vapor made using all the available long-term data sources (SAGE II, HALOE, balloon-borne instruments, ER-2-borne instruments) suggest that there are problems with most if not all of these measuring systems. Not only are the observed trends obtained with different instruments different, but there are significant differences in absolute abundance. These differences are not simply related to measurement type. The possibility of obtaining some very focused coincident measurements between in situ instruments aboard the WB-57 aircraft and those from SAGE II in the next few months was noted.
Both observational and modeling studies related to two questions on stratospheric water vapor were presented by Marvin Geller of the State University of New York at Stony Brook. These questions were the relationship between the annual variation of the hygropause and the cold-point tropopause temperature, and the factors giving rise to interannual variation in stratospheric water vapor. For the former, cold-point tropopause temperatures were obtained from spline fitting to daily European Center for Medium Range Weather Forecasting analyses and compared to temperature profiles from the U.S. National Centers for Environmental Prediction and also to radiosonde information from Kapingamarangi. Results show these cold points are predominantly located in the region of the atmosphere between 90° and 180° East and 20° S and N. The saturation mixing ratios associated with these cold-point tropopause temperatures were then calculated with the Clausius-Clapeyron equation, and shown to be significantly below those calculated using mean temperatures as has normally been done. The resulting water vapor mixing ratios are much closer to the minimum water vapor mixing ratios seen by SAGE, especially near their seasonal minimum. The results demonstrate clearly that tropical tropospheric air enters the stratosphere continuously through the locations with the coldest tropical cold-point tropopause temperature, which, given their spatial distribution, means that the tropical western Pacific is the main part of the "stratospheric fountain" throughout the year. Studies of the interannual variation in SAGE II water vapor observations were also carried out. Clear evidence was seen for a QBO effect, with increased upward motion during the easterly phase of the QBO. Examination of the difference between water vapor distributions in El Niño and La Niña periods gave a much-less-clear result. The relatively brief period of data analyzed and the complexity of overlapping El Niño and QBO effects made it difficult to conclusively demonstrate what the El Niño effects were. There was some evidence for a small negative trend in SAGE II water vapor concentrations in the lower stratosphere. This could be spurious, and should be analyzed further.
Several aspects of upper tropospheric and stratospheric water vapor were reviewed by Er-Woon Chiou of SAIC, Inc., including intercomparisons of water vapor data obtained from SAGE II, HALOE, and MLS, the seasonal variability of water vapor at northern midlatitudes, and the dryness of the tropical upper troposphere. The intercomparisons suggest a fairly complicated relationship between the three different water vapor data sets (relative agreement appears to vary with height and latitude), although these are complicated by the fact that the SAGE water vapor distributions were from the pre-Pinatubo period in which the water vapor distributions are usable, while the UARS (HALOE, MLS) observations were for the post-Pinatubo period. Seasonal variation studies showed that the dominant variation in lower stratospheric (26, 46 mb) water vapor in the tropical northern hemisphere was the annual cycle, while in the upper stratosphere (5, 10 mb) semi-annual variation dominated (in the tropical Southern Hemisphere upper stratosphere, the annual variation appeared to dominate as well). Empirical fits were developed to represent the water vapor composition of the upper troposphere; empirical scale heights of 1.2 - 1.6 km were obtained. Relative humidity values were calculated for the upper troposphere. Some relative humidity values as low as 5-10% were found; these are consistent with those obtained with recent reanalysis of other satellite instruments.
Factors controlling the distribution of upper tropospheric water vapor under clear-sky conditions were summarized by Minghua Zhang of SUNY Stony Brook. Several processes are importantsubsidence, advection of cirrus clouds, evaporation of precipitation, intermittent convection, large-scale eddies, and diffusion. In clear-sky regions there should be no internal moisture sources or sinks, however, so the moisture content should be dominated by horizontal and vertical advection, as interpreted through the slopes of air parcel trajectories. To help understand the origin of these slopes, a two-dimensional model was
developed to simulate atmospheric motions in a clear-sky region given lateral boundary conditions. The results show that subsidence does not necessarily mean dry conditionsit is the slopes of the trajectories that determine dryness, and these are determined by the lateral boundary conditions. The relationship between convection and moisture varies with closeness to convectioncloser to the convection, more convection increases moisture (a positive feedback), while farther away, the opposite is true. Whether or not the net feedback is positive or negative will depend on the spatial extent of the clear-sky region.
Water vapor climatologies derived from SAGE II data were reviewed by SAGE II Principal Investigator Pat McCormick of Hampton University. The first SAGE II water vapor climatologies covered three years, and have recently been extended to cover the full 5.3 years of pre-Pinatubo SAGE data. Key results to come from the climatologies include the observation that the lowest water vapor mixing ratios observed in the lower stratosphere occur in both hemispheres in the same season, suggesting a single source. Some seasonal differences do existfor example, in Northern Hemisphere winter, the hygropause is much better defined than in summer. There are significant differences in the water vapor distributions in the eastern and western hemisphere when 500-100 mb columnar water vapor is considered. Future work will involve separation of the 100-300 and 300-500 mb regions so that the effects of stratospheric air in the upper part of these regions at middle and higher latitudes can be examined.
Michael Newchurch of the University of Alabama in Huntsville discussed comparisons of SAGE ozone observations with those from Dobson Umkehr measurements using 8 Umkehr stations (6 in the Northern Hemisphere, 2 in the Southern Hemisphere). The Boulder station provided an example of the average results. By using coincident measurements to eliminate most geophysical variability, the difference between the Umkehr and the SAGE ozone trends in the upper stratosphere is approximately 0.05%/year for Boulder, but averaged over the larger number of Northern Hemisphere sites the difference is 0.2-0.3%/year. Differences for the Southern Hemisphere were slightly larger. This work is part of a broader scope of analyses of ozone trends comprising many researches the results of which will appear in the 1998 Ozone Trends Assessment Report (IOC/SPARC). That report will indicate that, based on sophisticated analyses of SAGE, Dobson Umkehr, and SBUV (/2) observations, the preponderance of evidence for upper-stratospheric ozone destruction is now incontrovertible.
Studies of interannual variability in SAGE ozone and nitrogen dioxide measurements, especially the presence of seasonal and QBO cycles, were presented by William Randel of the National Center for Atmospheric Research. These used the currently archived version 5.93 SAGE data. The SAGE data provide an excellent way to study the QBO effect on ozone, and show that some 2/3 of the signal is in the lower stratosphere with 1/3 in the upper stratosphere. The integrated signal is consistent with that obtained by TOMS, but the vertical dependence looks very different from that obtained with SBUV, which has much less vertical resolution than does SAGE. The same analysis was applied to NO2, and strong anti-correlation of the QBO effect with that for ozone was seen in the middle stratosphere because of the role NO2 plays in catalytically destroying ozone; in the lower stratosphere (below 26 km) they are positively correlated. The relationship of the two QBO cycles should provide a clear fingerprint that can be used for testing models. The rest of the variance was studied by doing an empirical orthogonal function (EOF) analysis, skipping the post-Pinatubo period. One interesting result is that there appear to be some discontinuities in the observed EOF contributions across the Pinatubo eruption period this is true for both ozone and NO2. Longer-term analysis of the NO2 data shows evidence for decreases over the time period 1984-1997, although the last few years (1993-1997) show increases; whether this is real is not yet understood.
The effects of aerosols on stratospheric chemistry were discussed by Hope Michelsen of AER, Inc. Her particular interest was the role of aerosols on interconversion of chlorine chemistry, especially the way in which the effective threshold temperature for repartitioning of chlorine is increased as aerosol loading is increased. Data from UARS as well as the flight of the Atmospheric Trace Molecule Spectroscopy (ATMOS) instrument aboard the Space Shuttle in 1993 were used for this work. Model results suggest existence of a hysteresis effectthat the rate of chemical reactions as temperature increases is slower than that while it is decreasing.
Results from climatological and modeling studies
carried out using SAGE data were presented by Matthew Hitchman of the University of Wisconsin. Several areas of work were summarized, including understanding seasonal zonal asymmetries in the SAGE constituents, the nature of the subtropical airmass boundary, and the use of models to simulate transport of SAGE constituents (the WISCAR 2D model) and the role of convection and gravity waves on constituent distributions (using the Wisconsin non-hydrostatic modeling system). The role of the Aleutian Anticyclone in affecting aerosol distributions was examined, and evidence exists to suggest that one can bring high-aerosol air up from North Africa into the northern Pacific, leading to maximum wintertime aerosols in the anticyclone (while aerosol amounts are reducing in the polar vortex). It appears that Africa represents a significant region in which air gets out of the tropics. In summertime, results suggest that the Indian monsoon may provide a mechanism for moving material from the Northern to the Southern Hemisphere. The results show that troposphere-stratosphere exchange appears to occur preferentially over the summertime oceans downstream of monsoon anticyclones, so in the northern summer it occurs in the western Pacific, while in the southern summer, it occurs in the south central Atlantic Ocean.
The uses of SAGE-observed quantities as input for climate models were presented by Andrew Lacis of the NASA Goddard Institute for Space Studies. The stratospheric heating by aerosol particles was shown to occur mainly by the thermal (infrared) contributions; the visible radiation impact was mainly scattering. The aerosol information was originally input into the climate model using column-averaged size distributions, but it seems apparent that one must do a better job, taking into account known evidence for altitude-dependent particle-size distributions. Future models will thus include an improved representation of aerosol properties, and should also have a more-refined treatment of water vapor in the stratosphere.
Studies of atmospheric chemistry using SAGE II data were presented by Ross Salawitch of the Jet Propulsion Laboratory. This work emphasized three areasvalidation of SAGE NO2 data, studying the effect of the Pinatubo eruption on NO2 values, and integrating SAGE observations with those from other sensors. Because of the known limitations of the SAGE NO2 data, work has focused on the last of these areas. The relationship between aerosol loading and distribution of NO2 and other nitrogen-containing constituents was studied by use of balloon data using the Mark IV interferometer, when data were available from September 1990, April 1993, and September 1993, corresponding to low (pre-Pinatubo), high, and moderate levels of aerosol loading. Results suggest that the basic picture of aerosol effects on nitrogen oxide repartitioning, especially the evidence of a "saturation" behavior in which there is some aerosol level beyond which greater amounts do not notably effect the partitioning, is correct.
Michael Newchurch summarized some work by David Rusch of the University of Colorado, in which an alternative approach for the inversion of SAGE data was developed. In this approach, the geometric inversion is done at each wavelength to obtain extinction vs. altitude profiles, after which Rayleigh scattering is subtracted, and aerosol contributions are then removed by assuming a lognormal aerosol size distribution. This contrasts with the Langley inversion, in which the order of separation and geometric inversion is reversed. Prior comparisons using a method developed at the University of Lille in France suggested that these two methods were quite similar, but the new results showed that differences in the lowest part of the stratosphere may not be negligible (the results are indistinguishable above 25 km). The Colorado algorithm appears to lead to ozone profiles that agree better with some midlatitude ozonesondes than is obtained with the Langley algorithm, but additional work is needed to look at whether the aerosol parameters obtained with the Colorado algorithm are in better or worse agreement with what is known about aerosol distributions. The Colorado method also propagates error bars in a way that better reflects the altitude dependence than does the current Langley algorithm.
The meeting was well attended, with lively discussions, and showed the many applications of the SAGE II data set.