----------------------------------------- Changes to the AIRS-RTA for PGE version 5 ----------------------------------------- Scott Hannon, 08 May 2007 Overview: The AIRS fast forward model used by the PGE version 5 (AIRS-RTA v5) is the same basic model as used in version 4, but with some algorithm changes and additions to improve the fast model's accuracy and increase its flexibility. The changes and additions can be summarized as: * improved tuning of the transmittances * improved frequencies for channels 1-130 * improved estimate of reflected downwelling radiance * new daytime radiance contribution for non-LTE emission * new vertical variability for CO2 profiles * new variable trace gases N2O, SO2, and HNO3 Each of these changes for v5 are discussed in more detail in the following sections. Transmittance Tuning: The transmittance tuning (applied as optical depth multipliers) used in v4 was based on a limited amount of validation data available at the time the fast model was finalized in January 2004. Over the next two years a much larger amount of validation data become available, and this has been used to improve the transmittance tuning in v5. A more detailed discussion of the validation data and the effects of tuning can be found in [Strow et al 2006]. For the most part the changes to tuning are relatively minor, affecting computed Brightness Temperature (BT) spectra at the few tenths of a Kelvin level or less. The main change is to "fixed" gases in the 15 and 4.3 um regions. Unlike v4, the v5 tuning was extended to stratospheric channels (based upon AIRS retrievals, as very little in situ validation data is available). The tuning of the 4.3 um channels took into consideration non-LTE effects. Some relatively minor changes were made to tuning in shortwave channels affected by N2O. The changes to water tuning are minor, the main changes being near 880 and 1320 cm^-1 where the old tuning appeared to have been influenced by HNO3. The CH4 tuning near 1305 cm-1 was adjusted by 2% on the recommendation of the retrieval team to reduces biases in retrieved CH4. Finally, a significant change has been made to ozone in the 10 um region. This tuning attempts to account for changes to ozone line parameters in the recent HITRAN 2004 databases versus the HITRAN 2000 upon with the v5 fast model is based. Channel Frequencies: An error in the frequencies of channels 1 to 130 (detector module 12) has been corrected for v5. The v5 channels have been shifted ~+1.5% of a channel width compared to v4. New fast model coefficients were generated for the shifted channels based upon revised model SRFs. No other channels are affected. The effects on BT spectra are on the order of a few tenths of a Kelvin or less. Reflected Downwelling Radiance: The algorithm used to estimate the radiance contribution from reflected downwelling atmospheric thermal emission has been totally redone for v5. The new algorithm required some restructuring of the radiative transfer code for computational efficiency, but it provides a more accurate estimate at little extra computational cost. Over ocean and other low reflectivity surfaces, the reflected downwelling therm is typically on the order of a few tenths of a Kelvin in the widow channels and negligible in non-window channels. However, over more reflective surfaces the radiance contribution can be much larger, and here the improved algorithm for v5 may help. For v5 the reflected downwelling thermal radiance contribution is first computed as a downward radiance at the surface in manner similar to the calculation of the main upward thermal radiance. This radiance is multiplied by a fudge factor "F" to adjust for the use of the same layer transmittances used by the upward radiance calculations, as these are not quite mathematically correct when used for downwelling radiances. The "F" factor is calculated using predictors based on the surface-to-space transmittance, the secant of the viewing angle, and the downward radiance, with coefficients determined by regression. The total reflected downwelling thermal radiance reaching AIRS is the the downward radiance at the surface times the surface reflectivity (times pi) and the surface-to-space transmittance. Non-LTE: New coefficients and code were added to model the effects of non-Local Thermodynamic Equilibrium (non-LTE) in the AIRS shortwave stratospheric channels during daytime. The effects of non-LTE emission are quite large; if not accounted for, obs-cal BT biases on the order of 10 Kelvin are typical. The non-LTE model included in the v5 AIRS-RTA is simple and sufficiently accurate to reduce daytime biases to similar levels as the nightime biases where non-LTE is not present. The non-LTE radiance contribution is computed as an independent term added to the standard (LTE) radiance. This non-LTE radiance is calculated using predictors based on the solar and satellite zenith angles as well the mean temperature in the top five AIRS layers, with coefficients determined by regression. For more details, see ref [De Souza-Machado et al 2006]. Variable CO2: The algorithm responsible for handling the variability of trace gas CO2 was modified to allow for more flexibility. In v4 the CO2 variability was limited to an overall profile scale factor, while in v5 the CO2 may be independently varied in each AIRS layer. This change will allow AIRS computations to make use of climatologically based CO2 profile shapes rather than being restricted to the default US Standard profile shape. Variable N2O, SO2, and HNO3: For v5 new code and coefficients has been added to allow for variability in trace gases N2O, SO2, and HNO3. These gases were previously "fixed" in the v4 and early AIRS fast models. The implementation of the variability for each of these gases is similar to how we handled the variability of CO2. None of these gases are currently retrieved in the v5 PGE. N2O (nitrous oxide) is fairly well mixed in the atmosphere and is only expected to vary by a a few percent around the mean. N2O absorption has a significant affect on AIRS radiances in a three spectra regions: moderately at 1250-1320, strongly at 2180-2250, and weakly at 2450-2600 cm^-1. The effects of variable N2O on AIRS BT spectra are estimated to be at the couple tenths of a Kelvin level. SO2 (sulfur dioxide) is typically present in the atmosphere at too low a concentration to be detectable to AIRS. However, volcanic eruptions often emit large amounts of SO2 and push it high into the tropopshere, and these volcanic plumes are often observable using AIRS. SO2 absorption affects AIRS radiances in three spectral regions: weakly at 1100-1137, strongly at 1330-1390, and very weakly at 2470-2520 cm^-1. The effects of variable SO2 on AIRS BT spectra can be anywere from zero to tens of Kelvins, depending on how much SO2 is present. HNO3 (nitric acid) is present in the atmsophere at a level at which the climatic and season variations are detectable by AIRS. Most of the HNO3 is located in the stratosphere, and it tends be at the low amounts in the tropics and larger amounts at high latitudes. HNO3 affects AIRS radiances with similarly strength in two spectra regions: the 850-920 cm^-1 window region, and the 1280-1350 cm^-1 lower/mid troposphere water region. The global variation in HNO3 results in change to AIRS BT spectra on the order of a few tenths of a Kelvin. --- references --- S. De Souza-Machado, L. L. Strow, S. E. Hannon, H. Motteler, M. Lopez-Puertas, B. Funke and D. P. Edwards "Fast Forward Radiative Transfer Modeling of 4.3 um Non-Local Thermodynamic Equilibrium effects for the Aqua/AIRS Infrared Temperature Sounder" Geophysical Review Letters, 34, L01802 (2006), doi:10.1029/2006GL026684 L. L. Strow, S. E. Hannon, S. De Souza-Machado, H. E. Motteler, and D. C. Tobin "Validation of the Atmospheric Infrared Sounder radiative transfer algorithm" Journal Of Geophysical Research, vol. 111, D09S06 (2006), doi:10.1029/2005JD006146 ---end of file---