Aerosol backscatter coefficients were calculated using multiwavelength aerosol extinction products from the SAGE II and III/ISS instruments (SAGE: Stratospheric Aerosol and Gas Experiment). The conversion methodology is presented, followed by an evaluation of the conversion algorithm's robustness. The SAGE-based backscatter products were compared to backscatter coefficients derived from ground-based lidar at three sites (Table Mountain Facility, Mauna Loa, and Observatoire de Haute-Provence). Further, the SAGE-derived lidar ratios were compared to values from previous balloon and theoretical studies. This evaluation includes the major eruption of Mt. Pinatubo in 1991, followed by the atmospherically quiescent period beginning in the late 1990s. Recommendations are made regarding the use of this method for evaluation of aerosol extinction profiles collected using the occultation method.

Stratospheric aerosol consists of submicron particles

The Stratospheric Aerosol and Gas Experiment (SAGE) is a series of
satellite-borne instruments that use the occultation method (both solar and
lunar light sourced) and have a lineage that spans 4 decades, originating
with the Stratospheric Aerosol Measurement II in 1978 (SAM-II,

Due to the SAGE instrument's level of precision and the limited aerosol number density in the stratosphere, validating the aerosol extinction products has proven challenging. Successful validation is further limited by the measured parameter itself since coincident stratospheric extinction measurements are scarce. Conversely, high-quality backscatter measurements from ground-based lidar instruments are more common and, despite operating at a fixed location, may provide sufficient coincident observations for an evaluation of the SAGE aerosol product. However, the backscatter and extinction coefficient products are not directly comparable.

Previous researchers have accomplished this comparison through the application of
conversion coefficients determined from balloon-borne optical particle counters
(OPCs; see

On the other hand,

Contrary to previous efforts to compare extinction and backscatter coefficients,
the extinction-to-backscatter (EBC) method proposed in this study required
relatively basic assumptions about the character of the underlying aerosol.
These assumptions include composition, particle shape, and the shape of the size
distribution (common assumptions in Mie theory, as further discussed below).
While combining Mie theory and extinction measurements to gain insight into the
nature of stratospheric aerosol is a common methodology
(e.g.,

The SAGE instruments used in the current study are SAGE II (October 1985–August 2005) and SAGE III on the International Space Station (SAGE III/ISS, June
2017–present, hereafter referred to as SAGE III). The SAGE II
instrument and algorithm (v7.0) have been described previously by

SAGE II was a seven-channel solar occultation instrument (386, 448, 452, 525,
600, 935, 1020 nm) that flew on the Earth Radiation Budget Satellite (ERBS) from
October 1984 through August 2005. Due to the orbital inclination and the method
of observation, SAGE II observations were limited to

Zonal overpass bar charts for SAGE II

SAGE III is a solar–lunar occultation instrument that is docked on the ISS and
has a data record beginning in June 2017. The onboard spectrometer is a charge-coupled device with a resolution of 1–2 nm. The spectrometer's spectral range
extends from 280 to 1040 nm in addition to a lone InGaAs photodiode at 1550 nm.
Similar to SAGE II, SAGE III has a higher frequency of observations at
midlatitudes compared to the tropics and high latitudes
(Fig.

Ground lidar data from three stations were used within this study. To allow
intercomparison with both SAGE II and SAGE III, candidate ground stations with a
long-duration data record were preferred. Further, data quality is likewise
important. The Network for Detection of Atmospheric Composition Change (NDACC,

The NASA Jet Propulsion Laboratory (JPL) Table Mountain Facility (TMF) is
located in southern California (34.4

The NOAA Mauna Loa Observatory (MLO; 19.5

The Observatoire de Haute-Provence (OHP; 43.9

Extinction and backscatter observations cannot be directly compared. In order
to evaluate the agreement between backscatter measurements and extinction
coefficient measurements, the data types must be converted to a common
parameter, thereby requiring a conversion algorithm. As previously mentioned,
this is usually done by converting backscatter to extinction coefficients using
conversion factors from sources independent of either instrument (e.g., constant
lidar ratio). Herein, we derive a process to infer this relationship based on
the spectral dependence of SAGE II/III aerosol extinction coefficient
measurements and only make basic assumptions on the character of the underlying
aerosol. Indeed, this EBC method is proposed to act as a bridge between aerosol
extinction and backscatter observations. This bridge is founded upon Mie theory

Particulate backscatter and extinction efficiency factors
(

At this point a technical note regarding construction of the lognormal
distribution must be made. Construction of a lognormal distribution fails when
the mode radius is near the limits of

Wavelengths were selected based on SAGE extinction channels and available lidar
wavelength, and the lookup tables were used to create the plots in Fig.

Theoretical relationship between the inverted lidar ratio
(

A potential limitation of this method is that, for large particle sizes
(extinction ratios

Extinction ratio PDFs (panels

While Fig.

Figure

To evaluate the robustness of the EBC algorithm,

Zonal statistics displaying the overall agreement between the two backscatter calculations using data collected during the SAGE II mission.
Panels

Same as in Fig.

Aggregate statistics for line-of-best-fit slope and

The zonal weighted coefficient of correlation (

Correlation plots (not shown) were generated for each latitude band and each
altitude from 12 to 34 km (2 km wide bins centered every 2 km) with corresponding
regression statistics to better understand how the agreement between the
backscatter products varied with altitude and latitude and to aid in defining
reasonable filtering criteria to mitigate the impact of spurious retrieval
products typically seen at lower and higher altitudes. We observed that data
collected between 15 and 31 km had higher coefficients of correlation, slopes
closer to 1, and a tighter grouping about the 1

As an evaluation of how much influence data outside the 15–31 km range had on
this analysis an ordinary line of best fit was calculated for each combination
of beta values (i.e.,

By considering only the mean slope and mean

From this evaluation we conclude that data outside 15–31 km significantly influenced the statistics and that the applicability of this conversion method is limited to regions where sufficient signal is received by the SAGE instruments, namely 15–31 km.

Having established an altitude range interval over which the EBC method remains
robust we can continue the evaluation of the aggregate statistics as shown in
Figs.

The high

In addition to comparing

From this evaluation we determined that the selection of extinction wavelength
combination had a minimal impact on the calculated backscatter products when
altitudes are limited to 15–31 km (i.e., each combination of SAGE wavelengths
yielded the same backscatter coefficient within the provided errors).
Therefore, we proceed with the current analysis by using the

As with any study that involves modeling PSDs, the dominant sources of
uncertainty are in the assumptions of aerosol composition and distribution
parameters. Here, the particle number density and mode radius play a minor
role. However, as seen in Fig.

Another challenge in comparing SAGE and lidar observations is the differing
viewing geometries. The uncertainty introduced by these differing geometries
cannot be easily accounted for. However, current versions of the algorithm

The EBC method was applied to SAGE II and SAGE III datasets for intercomparison with ground-based lidar products. A discussion of the results of each SAGE mission follows.

The SAGE II record spanned over 20 years and had the benefit of observing the impact of two of the largest volcanic eruptions of the 20th century: recovery from El Chichón in 1982 and the full life cycle of the Mount Pinatubo eruption of 1991, followed by a return to quiescent conditions in the late 1990s. Within this record the extinction and backscatter coefficients spanned nearly 2 orders of magnitude, providing an interesting case study.

Time series of SAGE II (monthly zonal mean), lidar (monthly mean)
backscatter coefficients, and a SAGE-based lidar ratio estimate at 355 nm (monthly zonal mean) over the Table Mountain Facility. The spread in the
SAGE-derived backscatter and lidar ratios (both coefficients at same
wavelength) represents the range of values due to changing the spread
(

Same as Fig.

Same as Fig.

SAGE II data were used to estimate

Intercomparison statistics for the time series in
Figs.

Statistics for the time series data are presented in Table

Data collected at 15 km showed the worst agreement due to atmospheric opacity
and cloud contamination as discussed above. Conversely, the agreement was best
at 20 and 25 km (percent difference within

It was observed that during the Pinatubo time period the coefficients of correlation and line-of-best-fit slopes were higher than during background conditions. This was expected behavior for background conditions for two reasons: (1) in the absence of stratospheric injections the instruments were left to sample the natural stratospheric variability (similar to noise), which limits correlative analysis outside long-duration climatological trend studies, and (2) the limited dynamic range of the observations essentially provides a correlation between two parallel lines. Overall, the percent differences for TMO show the two techniques to be in good agreement, with worse agreement occurring at 15 km, which was expected due to cloud contamination.

Unlike TMO, the OHP lidar record did not start until

In addition to

Similar to OHP, the MLO record did not begin until

The statistics in Table

To date, the SAGE III mission has made observations under relatively clean stratospheric conditions similar to conditions at the end of the SAGE II mission. Due to the limited data record (3 years since launch), the comparison between SAGE III and the Mauna Loa and OHP lidars will be cursory. Data from the Table Mountain Facility have not been released for this time period; therefore, Table Mountain was excluded from the current analysis.

Same as Fig.

Same as Fig.

The SAGE III and lidar backscatter coefficients show similar qualitative
agreement at both Mauna Loa and OHP (Figs.

Similar to the end of the SAGE II record, calculation of a meaningful

Same as Table

For the SAGE II instrument the derived

Perhaps the most striking feature of this analysis is how well the SAGE-derived
backscatter coefficient agreed with the lidar record during the early stages of
the Pinatubo eruption (Fig.

The calculated

A method of converting SAGE extinction ratios to backscatter coefficient
(

A major finding of this research was the demonstration of the robustness of the
conversion method. It was shown that, within the specified error bars, the
calculation of

The robustness of the conversion method provides an indirect validation of the SAGE aerosol spectra. If the EBC method were wavelength dependent, this would indicate a substantial error in the standard aerosol products. However, our evaluation showed that the EBC is not wavelength dependent, thereby lending credence to the SAGE aerosol product wavelength assignment.

It was shown that, overall, the SAGE II-derived

For the SAGE III analysis only OHP and MLO were available for comparison. The
SAGE III-derived

A potential application of this method is informing lidar ratio (

Another application of this method may be providing global backscatter profiles independent of a space-based lidar such as CALIPSO. While we do not suggest that SAGE-derived backscatter coefficients can replace lidar observations, our product may be a viable alternative. With CALIPSO scheduled for decommissioning no later than 2023 (Mark Vaughan, personal communication, 2020) and no replacement scheduled for flight prior to its decommissioning date, the SAGE III backscatter product may provide a necessary link between CALIPSO and the next space-based lidar to ensure continuity of the record and provide a method of evaluating the performance of the next-generation orbiting lidar in the context of the SAGE III record and, by association, CALISPO.

SAGE data used within this study are available at NASA's
Atmospheric Science Data Center (

TNK and LT developed the methodology, while TNK carried out the analysis, wrote the analysis code, and wrote the paper. TL and FC provided lidar data collected at TMF and MLO and assisted in the description of this data product in the paper. SK and SGB provided lidar data collected over OHP and assisted in the description of this data product in the paper. MR, RD, KL, and DF participated in scientific discussions and provided guidance throughout the study. All authors reviewed the paper during the preparation process.

The authors declare that they have no conflict of interest.

This article is part of the special issue “New developments in atmospheric Limb measurements: Instruments, Methods and science applications”. It is a result of the 10th international limb workshop, Greifswald, Germany, 4–7 June 2019.

SAGE III/ISS is a NASA Langley managed mission funded by the NASA Science Mission Directorate within the Earth Systematic Mission Program. The enabling partners are the NASA Human Exploration and Operations Mission Directorate, the International Space Station Program, and the European Space Agency. SSAI personnel are supported through the STARSS III contract NNL16AA05C.

This research has been supported by the NASA Science Mission Directorate within the Earth Systematic Mission Program, the NASA Human Exploration and Operations Mission Directorate, the International Space Station Program, the European Space Agency, and STARSS III (grant no. NNL16AA05C).

This paper was edited by Chris McLinden and reviewed by three anonymous referees.