Last update: 24-Oct-2003

A new version of the SAGE II data products has been released. The primary change to the algorithm dealt with the improvement in the water vapor product. The SAGE II V6.1 data has not been publically available after mid-2000 due to an altitude registration problem. This has been tracked down and corrected. The release of V6.2 contains all data through September 2003 and subsequent months will be added on approximately the middle of the following month. An error in the interpolation of the NCEP Met data used to remove Rayleigh scattering from the transmission profiles has also been corrected. The data are currently available from our server and will be available through the EOSDIS system after Oct 28, 2003.

Data since July 2000

In the middle of July 2000, SAGE II experienced a failure of its azimuth gimbal system. This was corrected and operation at a 50% duty cycle was reestablished by November 2000. A few months after the return to operation it became apparent that there were sporadic problems in quality of the data products. Many potential sources (Y2K, Instrumental, ephemeris epoch, and operational interferences) were suspected. After a lengthy investigation it appeared that the problem periods followed a peculiar pattern of an abrupt change in quality at midnight GMT on Sundays and Thursdays. It was also observed that the problem was consistent with an altitude registration problem of up to 3 km, or more in some cases. Working with the flight operations and flight dynamics folks it was determined that the problem was due to a change in the software used to produce the final satellite ephemeris from the tracking data. The change-over occurred at essentially the same time as the gimbal failure. The software has been changed to closely match the performance of the pre-2000 software and has been used to produce a new satellite ephemeris data set from Jan 1, 2000 onward. The data after mid-2000 now exhibit the proper behavior.

V6.1 has achieved a significant improvement in the NO₂ retrievals through a minor modification to the spectroscopy. The locations of both the 448 and 452 nm channels were adjusted slightly to compensate for an obvious problem in Rayleigh subtraction due to improper spacing of the channels. In addition, the temporal broadening of the 448 nm channel filter was shifted to a narrower initial FWHM. As in V6.0, the evolution of this filter width follows the observed increase in throughput relative to the 452 nm channel. Additionally, the sunrise 448 and 452 nm transmission profiles were affected by reversing the order of the 'electronic transient' correction and the scan mirror calibration. The NO₂ corrections cause a concomitant improvement to short wavelength aerosol extinction particularly in the lower stratosphere and during low aerosol loading periods. It also has an impact on the detection of polar mesospheric clouds.

The ozone data product was also improved through an adjustment to aerosol clearing and the correction of channels at 525 and 1020 nm for absorption by the oxygen dimer. Several new products are introduced in this version. Aerosol surface area density, computed using a method similar to that described in Thomason et al. [J. Geophys. Res., 102, 8967-8976, 1997] and aerosol effective radius are now incorporated directly into the primary data files. A new middle atmosphere density product is calculated using only SAGE II measurements, independent of any outside temperature and pressure data set. Rayleigh (or molecular) contribution to the measurements is computed in a manner similar to the regular species separation but assumes no aerosol contribution. The results are calculated and reported from 40.5 km to 75.0 km. Preliminary assessment suggests that the behavior is reasonable from approximately 42 km to 58 km, though noisy on an event-by-event basis. With the new data products, the file format of the species files has changed from the previous version. IDL and Fortran 90 include files for the new data format are available at the ftp. Also available are IDL-based data viewers.

Water Vapor

The SAGE II water vapor retrieval has been significantly updated in this version. Version 6.1 exhibited a significant dry bias that was as large as 2 ppm in the vicinity of the hygropause. Since water vapor is never more than 10% of the total signal in the nominal "water vapor channel", the commonly held view is that the problems in water vapor are related to clearing interfering species including molecular scatter, ozone, and particularly aerosol. During a critical evaluation of the aerosol model used in the water vapor algorithm, it became apparent that it was not possible for aerosol to produce the observed bias in the water vapor product. Further examination strongly suggested that, like the filter-based NO₂ channels, the filter-based water vapor channel had significantly drifted in spectral response particularly in the first year of operation. While it is difficult to determine the details of the modified response, we have forensically deduced that the channel has moved approximately 10 nm toward longer wavelengths and widened slightly. With this new location, the Version 6.2 water vapor product, at least in the absence of significant volcanic aerosol perturbations, no longer exhibits the dry bias and matches the HALOE water vapor climatology well from 15 to 30 km (slightly wetter at the upper end of this range).

Click here to view the JGR manuscript of the paper entitled
"A Revised Water Vapor Product for the SAGE II Version 6.2 Data Set"

Water Vapor Caveats

Given the drift in the spectral response and the limited ability to fit it's time dependence, it is quite likely that any trend derived from this data would reflect a trend in the goodness of the fit as much as an actual trend in water vapor itself.

Since the bulk of the spectral shift occurred in 1985, we have not developed a time-dependent model for this period and caution should be used in making use of this data. In 1984 and early 1985, retrievals using the nominal channel location look reasonable. Version 6.2 uses the nominal location for 1984 and 1985 and an abrupt shift to the new channel location beginning in 1986.

Water vapor is very sensitive to aerosol loading and, based on sensitivity studies, we do not recommend using SAGE II water vapor in the stratosphere when the associated 1020-nm aerosol extinction exceeds 4. 10-4 km-1. This criteria eliminates at least some data in 1985-1986 and again in 1991-1994. Other periods are mostly unaffected by this limitation.

Ozone

For the most part, the ozone density profiles have changed on the order of 0.5% from version 6.1. The changes may be larger above 50 km and are due primarily to the correction of an altitude registration problem in our NCEP gridding algorithm. The increase in the retrieved water vapor in the troposphere results in even lower ozone values below 10 km.

Bit Flag Changes

In version 6.20, the meanings of some of the bit flags in the two quality vectors have changed. These vectors are the elements called "infvec" in the index and species structure files. The complete set of bit flag descriptors is available in the IDL structure file "bitinfo_620.pro"

Three new bit flags have been added to the quality vector in the index file, relating to changes in the water vapor processing. These are bits 5, 6, and 7, all of which were spare bits in version 6.10. Bits 5 and 6 are used to distinguish aerosol clearing models for the water vapor channel. If bit 5 is on, the model is the El Chichon model. If bit 6 is on, the model is the Pinatubo model. Both bits 5 and 6 are turned on during a transition period between the two models. If neither of these flags is turned on, the model is the same as in version 6.10. Bit 7 serves as a signal that the filter function for the water vapor channel was changed from earlier versions. This bit is on for all events after 1985.

Changes have also been made to the bit flags in the altitude dependent information vector, found in the species structure file. Bits 11 and 12 in version 6.10 were the "large 525 OD" bit and the "large 1020 OD" bit that were set when the total optical depth in those channels becomes very large, possibly signaling the presense of a cloud. In the current version, the "Large 1020 OD" bit has been moved to bit 5 (formally a spare) and bits 11 and 12 are replaced with the result of a cloud test [Kent, G. S., P.-H. Wang, M. P. McCormick, and K. M. Skeens, Multiyear SAGE II measurements of upper tropospheric aerosol characteristics, J. Geophys. Res., 100, 13,875- 13,899, 1995]. Testing for clouds occurs starting in 1985 between 6 km and 25.5 km altitude. If the test is performed and the result is well determined, then bit 11 is turned on, and bit 12 shows whether a cloud is present (on) or absent (off). If bit 11 is off, bit 12 shows whether the test was indeterminate (on) or not performed (off).

     Click here to access the data using your web browser

Anonymous FTP is also available:

% ftp ftp-rab.larc.nasa.gov
Connected to ftp-rab.larc.nasa.gov.
Name (ftp-rab.larc.nasa.gov:anonymous): anonymous
331 Guest login ok, send your complete e-mail address as password.
Password: ********
230 Guest login ok, access restrictions apply.
ftp> cd /pub/sage2/v6.20
250 CWD command successful.
ftp> mget *
.
.
(your session)
.
.