The derivation of the Triana total ozone amounts is based on the TOMS (Total Ozone Mapping Spectrometer) algorithms with adjustments for the Triana view angles. Four of the EPIC UV wavelengths (317.5± 0.5, 325± 0.5, 340± 1.5, and 388± 1.5 nm) were chosen to match closely those that were used by the highly successful original Nimbus-7/TOMS instrument (1978 to 1993). The fifth UV wavelength is centered on the solar calcium-K Fraunhofer line at 393.5± 0.5 nm. Filling in of the Fraunhofer line as a function of altitude is used for cloud height analysis (see discussion below) and to improve the retrieval of total column ozone.
The amount and distribution of total ozone over the globe is sensitive to the state of the atmosphere with regard to pollution (e.g. man-made chlorine bearing chemicals) and the effects of atmospheric temperature changes. The total amount of ozone contained in a column is obtained from the ratios of measured radiances I(317.5)/I(340) or I(325)/I(340). The reduced sensitivity of 325 nm to ozone absorption compared to 317.5 nm is used to extend the measurements to higher solar zenith angles than is possible with 317.5 nm. At high solar zenith angles, the 317.5 nm solar irradiance does not penetrate all the way to the surface, and so does not detect the total column ozone amount. The radiance at 340 nm is almost unaffected by ozone absorption, and is used as the reference channel to characterize the Rayleigh scattering atmosphere. The 388 nm channel can also be used as a reference channel.
The method of inversion to obtain ozone amounts from the measured radiances is based on precomputed lookup tables derived from radiative transfer solutions. The algorithm includes the effects of clouds derived from a measurement of the increased scene reflectivity (340 or 388 nm) over the normal clear-sky UV surface reflectivity (2 to 8%). Corrections are also made for the presence of aerosols within each scene (dust, smoke, and pollution, see below).
A further measure of ozone can be obtained using the weak absorption in the Chappuis band at 551 nm as discussed in the following paragraphs. An example, shown in Figure 12, of the expected ozone detection capability has been simulated using data from TOMS.
Chappuis-band ozone detection is used to extend the latitude range over which measurements can be accurately made (to within 5%). This will allow EPIC to observe the development of ozone changes in the sunlit portion of the Arctic, particularly during the important spring season (see Figure 13). Depending on the orbit, EPIC will also be able to observe the springtime development of the Antarctic ozone hole.
For radiation at 317.5 or 325 nm, only a small fraction of the photons back-scattered from the atmosphere come from low altitudes when solar zenith angles are large (SZA > 70°). The problem arises from two sources: both the ozone absorption and Rayleigh scattering are roughly proportional to e EXP(-(a n+ß N)/2Cos(SZA)) for EPIC observations,
where a is the ozone absorption coefficient (cm¯¹)
n is the column amount of ozone (cm)
ß is the Rayleigh scattering coefficient (cm¯¹)
N is the column amount of molecular atmosphere (cm).
The result is that UV wavelengths that are weakly sensitive to ozone absorption cannot be used at high SZA because of intense Rayleigh scattering. The problem is made worse as N/Cos(SZA) increases because of multiple scattering effects. The Rayleigh scattering problem can be greatly reduced if measurements are made in the visible wavelengths where there is also weak ozone absorption.
The peak Chappuis-band ozone absorption in the visible wavelengths occurs near 605 nm and is negligible for wavelengths shorter than 450 nm and longer than 750 nm. As currently configured, EPIC contains a filter position at 551± 5 nm (green) where the Chappuis band ozone absorption is still strong and where the Rayleigh scattering is relatively small. The reference channel could be one of the following existing wavelength channels, 443, 645, or 870 nm. Radiative transfer analysis indicates that 443± 5 nm (blue) is the best choice, since it has almost no ozone absorption (compared to 645 nm) and is much closer to 551 nm than 870 nm. The more sensitive channel at 605 nm was not used so as to include a water-sensitive channel at 905 nm and still have only 10 wavelengths.
As with other calculations, the radiative transfer analysis has been performed with a full spherical geometry calculation (Herman et al., 1996) and with the pseudo-spherical program that has been extensively validated over the past 20 years (Dave, 1965). Both calculations agree up to 80° SZA with the results from the full spherical geometry calculation used between 80° and 90°. The results are contained in a lookup table for C(ozone, SZA).
C(ozone, SZA) = I443 / I551
At solar zenith angles near 60°, where total column ozone can be determined by both I443/I551 and I325/I340, the values will be compared to assess the accuracy of the Chappuis-band analysis. This is needed because the Chappuis-band estimation of ozone is sensitive to the underlying surface reflectivity, which is variable in the blue and green wavelengths. The blue and green surface reflectivities will be estimated at smaller SZA and used at angles greater than or equal to 60°. A possible problem is that the surface reflectivities have an angle dependence that is not known for Triana observing conditions, and can cause an error in calculated ozone amounts. The comparison with the I325/I340 determination of ozone will help determine this angular dependence.
