1. TAFTS: Atmospheric Profile Uncertainty and Continuum Contribution Ralph Beeby Paul Green, Juliet Pickering 29 th September 2010

2. Outline Introduction Atmospheric profiles Uncertainty in profiles; calculating equivalent uncertainty in radiance Comparison: uncertainty in profiles from B400 against simulated changes in continuum strength Conclusions

3. Introduction TAFTS (Tropospheric Airborne Fourier Transform Spectrometer) used to measure in- situ far-infrared atmospheric radiances during CAVIAR field campaigns Aim: to use TAFTS measurements to measure the water vapour continuum as part of CAVIAR Need to compare TAFTS spectra with simulations based on existing line databases LBLRTM (Line by Line Radiative Transfer Model): - can easily adjust strength of continuum absorption within this model - can use real atmospheric measurements as input to model - includes Analytic Jacobian routine – calculating sensitivity of simulation to uncertainties in profile Would a change in continuum absorption be greater or smaller than the change in radiance due to uncertainties in profile?

4. Atmospheric Profiles LBLRTM takes a 1D atmospheric profile as input Radiosondes, dropsondes and ECMWF data used to generate a best estimate of profile for Camborne flights Basic measurement uncertainty of ±2%RH and ±0.2 o C from sonde measurements Additional uncertainty due to separation of sondes from aircraft, i.e., Separation of sonde from aircraft Variation of profile with time and space Representation Error Other instruments onboard FAAM aircraft can be used to measure temperature and water vapour content

5. Variability of Atmospheric Profiles 2 x Rosemount Temperature Probes (T), General Eastern Dewpoint Hygrometer (RH) How does relative humidity vary over the course of a level run? Have hygrometer data for five levels in model atmosphere – what about the other 144? Standard deviation of RH appears to be roughly proportional to magnitude RH in profile – use this to extrapolate standard deviation for all levels in profile Have since calculated similar standard deviations for all flights in Camborne campaign – correlation seems to hold!

6. Calculating Equivalent Uncertainty Analytic Jacobian: calculates R/x for each level and each wavelength, where R is radiance and x is the parameter of interest (RH or T in this case) Indicates how sensitive the spectrum will be to a given change in parameter x at each level So to calculate error: multiply R/x by randomised RH/T uncertainty at each level, sum contribution from all levels between boundary (surface or TOA) and observer Compare with LBLRTM spectra in which the foreign continuum absorption strength has been adjusted Wavenumber / cm -1 dR/d log[vmr(H2O)] / mWm 2 sr.cm -1 /log[vmr]

7. Comparison: Upwelling Radiance

8. Comparison: Downwelling Radiance

9. Conclusions Compared uncertainty in atmospheric profiles with uncertainty in continuum strength Equivalent change in radiance simulated using LBLRTM (continuum) and calculated from Jacobian routine using flight data (profile) Longer wavelengths less sensitive both to profile uncertainty and changes in continuum in terms of radiance Profile uncertainty more significant in drier region of atmosphere around 20,000ft at higher wavenumbers (shorter wavelengths) Continuum contribution more sensitive in wetter regions, particularly for downwelling radiation Use this as a guide to which TAFTS data to use to look for continuum signal, e.g., 34,000ft run, around 238cm -1 downwelling, 5,500ft around 365cm -1 downwelling