V8 Algorithm Description

The Solar Backscatter Ultraviolet instruments, SBUV on Nimbus 7 and SBUV/2s on NOAA-9, -11, -14, -16 and -17, are nadir-viewing instruments that infer total column ozone and the ozone vertical profile by measuring sunlight scattered from the atmosphere in the ultraviolet. Heath et al. (1975) describes the SBUV flown on Nimbus-7. Frederick et al. (1986), and Hilsenrath et al. (1995) describe the follow-on SBUV/2 instruments flown on the NOAA series of spacecraft.

The instruments are all of similar design: nadir viewing double grating monochromators of the Ebert-Fastie type. The instruments step through 12 wavelengths in sequence over 24 seconds, while viewing the Earth in the fixed nadir direction with an instantaneous field of view (IFOV) on the ground of approximately 180 by 180 km. To account for the change in the scene-reflectivity due to the motion of the satellite during the course of a scan, a separate co-aligned filter photometer (centered at 343 nm on SBUV; 380 nm on SBUV/2) makes 12 measurements concurrent with the 12 monochromator measurements. Each sequence of measurements is separated by 8 seconds from the next producing a complete set every 32 seconds on the daylight portions of an orbit.

The instruments are flown in polar orbits to obtain global coverage. Since the SBUV ozone measurements rely on backscattered solar radiation, data are only taken on the dayside of each orbit. There are about 14 orbits per day with 26° of separation at the equator. Unfortunately, the early NOAA polar orbiting satellites are not sun-synchronous. For example the NOAA-11 equator crossing times drifted from 1:30 pm (measurements at 30° solar zenith angle at the equator) at the beginning of 1989 to 5:00 pm by the end of 1995 (measurements at 70° solar zenith angle). As the orbit drifts, the terminator crossing location moves to lower latitudes and coverage decreases.

Ozone profiles and total column amounts are derived from the ratio of the observed backscattered spectral radiance to the incoming solar spectral irradiance, which is referred to as the backscattered albedo. The only difference in the optical components between the radiance and irradiance observations is the instrument diffuser used to make the solar irradiance measurement; the remaining optical components are identical. Therefore, a change in the diffuser reflectivity will result in an apparent trend in ozone. This is the key calibration component for the SBUV(/2) series. See Hilsenrath et al. 1995 for a longer discussion.

The spectral resolutions for SBUV(/2) monochromators are all approximately 1.1 nm, FWHM with triangular bandpasses. The bandwidth of the photometer is about 3 nm. The wavelength channels used for Nimbus 7 SBUV were: 256, 273, 283, 288, 292, 298, 302, 306, 312, 318, 331, and 340 nm. The wavelengths for NOAA-9 and the other NOAA SBUV/2 instruments were very similar except that the shortest channel was moved from 256 nm to 252 nm in order to avoid emission in the nitric oxide gamma band that contaminated the SBUV Channel 1 measurement. Data from the 256 and 252 nm channels are not used in this processing for any of the instruments.

Version 8 SBUV(/2) Ozone Profile Retrieval Algorithm (V8A)

The Version 8 SBUV(/2) ozone profile retrieval algorithm combines backscattered ultraviolet measurements and a priori profile information in a maximum likelihood retrieval. (See Rodgers 1990 for an analysis of this class of retrievals.) It improves on the Version 6 SBUV(/2) algorithm described in Bhartia et al. 1996. Among the improvements are the following:

  • 1. The V8A has new set of a priori profiles varying by month and latitude leading to better estimates in the troposphere (where SBUV/2 lacks retrieval information) and allowing simplified comparisons of SBUV/2 results to other measurement systems (in particular, to Umkehr ground-based ozone profile retrievals which use the same a priori data set).
  • 2. The V8a has a true separation of the a priori and first guess. This simplifies averaging kernel analysis. Examples and further information are provided below.
  • 3. The V8A has improved multiple scattering and cloud and reflectivity modeling. These corrections are updated as the algorithm iterates toward a solution.
  • 4. Some errors present in the V6A will be reduced. These include the elimination of errors on the order of 0.5% by improved fidelity in the bandpass modeling.
  • 5. The V8A incorporates several ad hoc Version 6 algorithm improvements directly. These include better modeling of the effects of the gravity gradient, better representation of atmospheric temperature influences on ozone absorption and corrections for wavelength grating position errors.
  • 6. The algorithm uses improved terrain height information and gives profiles relative to the estimated surface pressure.
  • 7. The V8A is also designed to allow the use of more accurate external and climatological data and allow simpler adjustments for changes in wavelength selection.
  • 8. Finally, the V8A is designed for expansion to perform retrievals for hyperspectral instruments, such as OMI, GOME-2 and the Nadir Profiler in OMPS.

    The atmospheric ozone absorption decreases by several orders of magnitude over the 252 to 340 nm wavelength range. The V8A uses a variable number of backscattered ultraviolet measurements depending on the solar zenith angle (SZA) of the observations to maintain its sensitivity to ozone changes in the lower atmosphere. For small solar zenith angles (the sun high in the sky) only six wavelengths are used in the retrievals. They are at 273 nm, 283 nm, 288 nm, 292 nm, 298 nm, 302 nm and 306 nm. As the SZA increases the 306 nm, then 313 nm and finally the 318 nm channels are added to the retrieval.

    V8 A priori Profiles

    The a priori profiles are provided in a data file (climat.txt). These profiles can be used to determine the information used in a specific retrieval by interpolating in latitude and day with a FORTRAN code. The lowest layer from the a priori program will differ from that used in the retrieval if the surface pressure is not 1 atm. Another data file (terrain.txt) and FORTRAN code are provided to generate the surface pressure for a given latitude and longitude.

    The a priori covariance matrix is constructed as follows: the diagonal elements correspond to 50% error and the non-diagonal elements fall off with a correlation length of 12 fine layers (approximately two Umkehr layers). The measurement covariance is diagonal and corresponds to radiance errors of 1% in each channel.

    The profile database was created from 15 years (1988-2002) of ozonesonde measurements and SAGE (Version 6.1) and/or UARS-MLS (Version 5). Over 23,400 sondes from 1988-2002 were used in producing this climatology. Data was "filtered", i.e., obvious bad data points were removed. Balloons that burst below 250 mbar are discarded. Data from bouncing balloons were sorted by pressure. Note: No total ozone correction factor (TOMS or Dobson) filtering was used. The stations were weighted equally for each band so that we do not introduce any longitudinal biases (e.g. Resolute & Nyalesund have equal weights in December even though Nyalesund has 3 times as many sondes than Resolute for that month). The SAGE data was also "filtered" for bad retrievals. Average profiles from ozonesondes and SAGE are merged over a 4-km range with the sonde weight decreasing from 80 to 60 to 40 to 20% and the SAGE weight increasing correspondingly.

    Averaging kernel, Retrieval noise, Degrees of freedom, and Resolution plots for Nimbus 7 (1979-03-20)

    Fig. 1 shows Averaging Kernels (AK) (for fractional changes in ozone) at the 15 pressure levels where the ozone mixing ratios are provided on this DVD. The short horizontal lines on the right side of the graph shows the pressure levels, and the line style points to the corresponding AKs. In general, the (fractional) variation in the mixing ratio reported by SBUV at a given pressure level is a weighted average of the (fractional) variation of the mixing ratio at surrounding altitudes, with respect to the apriori profile. Since the SBUV V8 apriori profiles have no inter-annual variation, the AKs also show how the algorithm smoothes the long-term trends in ozone mixing ratio.Note, however, that individual SBUV profiles usually have structures that are finer than those implied by the AKs; these structures come from the assumed apriori profile, rather than from the measurements themselves. This figure shows typical AKs at the equator. The AKs show best resolution of ~6 km near 3 hPa, degrading to ~10 km at 1 and 20 hPa. Outside this range the retrieved profiles have little information. For example, the (fractional) variation in ozone mixing ratio seen at 0.5 hPa actually represents the (fractional) variation from the region around 1 hPa, and the variations around 50 hPa represents the variations from around 30 hPa.

    Fig. 2 shows typical AKs for March at 40N latitude. At this latitude the 50 hPa AK does capture the atmospheric variation, albeit with a resolution of ~11 km. In general, the upper AKs get progressively better as the solar zenith angle increases, and the lower AKs become better as the ozone density peak drops in altitude.

    Fig. 3 shows typical AKs for March at 80N latitude. It exemplifies the improvement in the low hPa AKs, especially the 0.5, 0.7, 1, and 1.5 hPa AKs, in capturing the atmospheric variation more closely than in Figs. 2 and 1. There is also an improvement in resolution at the low hPa end, to ~6 km up to 0.8 hPa.

    References

    Bhartia, P.K., S. Taylor, R.D. McPeters, and C. Wellemeyer, Application of the Langley plot method to the calibration of the solar backscatter ultraviolet instrument on the Nimbus 7 satellite, J. Geophys. Res. 100, 2997-3004, 1995.

    Bhartia, P.K., R.D. McPeters, C.L. Mateer, L.E. Flynn, and C.G. Wellemeyer, Algorithm for the estimation of vertical profiles from the backscattered ultraviolet technique, J. Geophys. Res. 101, 18,793-18,806, 1996.

    Frederick, J.E, R. P. Cebula, and D. F. Heath, Instrument characterization for the detection of long-term changes in stratospheric ozone: An analysis of the SBUV/2 radiometer, J. Atmos. Oceanic Technol., 3, 472-480, 1986.

    Gleason, J.F, R.D. McPeters, Correction to the Nimbus 7 solar backscatter ultraviolet data in the "nonsync" period (February 1987 to June 1990), J. Geophys. Res., 100, 16,873-16,877, 1995.

    Heath, D. F., A. J. Krueger, H. R. Roeder, B. D. Henderson, The solar backscatter ultraviolet and total ozone mapping spectrometer (SBUV/TOMS) for Nimbus G, Optical Engineering, Vol. 14, 323-331, 1975.

    Heath, D.F., Z. Wei, W.K. Fowler, and V.W. Nelson, ''Comparison of Spectral Radiance Calibrations of SSBUV-2 Satellite Ozone Monitoring Instruments using Integrating Sphere and Flat-Plate Diffuser Technique,'' Metrologia, 30, 259-264, 1993.

    Hilsenrath, E., R.P. Cebula, M.T. Deland, K. Laamann, S. Taylor, C. Wellemeyer, and P.K. Bhartia, Calibration of the NOAA-11 Solar Backscatter Ultraviolet (SBUV/2) Ozone Data Set from 1989 to 1993 using In-Flight Calibration Data and SSBUV, J. Geophys. Res., 100, 1351-1366, 1995.

    Rodgers, C. D., The Characterization and Error Analysis of Profiles Retrieved from Remote Sounding Measurements, J. Geophys. Res., 95, p5587-5595, 1990.

    Examples of a Data Filter

    The error codes in the data files are determined by using a single set of measurements. The ResQC values are also computed for individual profiles. Comparing the results to nearby values can also disclose anomalous behavior. In particular, the effects of noise spikes in the shortest wavelength channel (273 nm) are difficult to detect for individual retrievals, but comparing the upper layers to those for adjacent retrievals will often identify suspect profiles. We give an example of a screening procedure to help to determine if there is a problem with some profiles. Like any screening procedure there is a danger that interesting real profile may be eliminated along with those produced by unusually noisy measurements. A sample IDL code, which has proven useful in identifying "bad" profiles, is provided. It uses the product of the deviation of the upper stratospheric mixing ratios from those in adjacent retrieval profiles as a measure of unusual behavior. This screen will identify 0.01% of the data as bad during periods of good performance for the lowest noise instruments (NIMBUS-7 prior to 1987 and NOAA-9). For other instruments or during periods of poorer performance it may identify as much as 3% of the data as being of suspect quality due to effects associated with noisy measurements. The screen will sometimes flag profiles as suspect when they are normal but are adjacent to unusual profiles.