Comparing high resolution far and mid infrared spectra for clear – sky atmospheric profile retrievals

High spectral resolution observations of the atmospheric emission spectrum in the infrared have proven utility for atmospheric thermodynamic state retrievals. Current and future operational satellite instruments (AIRS, IASI, CrIS) produce these measurements in the Mid Infrared (MIR) portion of the spectrum (roughly 4 – 15 μm) on a routine basis. In comparison, the Far Infrared (FIR) portion of the spectrum (wavelengths longer than 15 μm) is poorly observed, mainly due to sensor limitations in this wavelength region.

In this study we are investigating the differences between retrievals based on MIR and FIR measurements, to understand how FIR measurements will improve our ability to sense the atmosphere. Water vapor, liquid water, and ice all have different absorption properties in the MIR and FIR spectral regions. For example, the FIR contains the rotational absorption band of water vapor, with no significant interference from minor atmospheric constituents, while the MIR contains the vibrational – rotational absorption band of water vapor, along with interference from minor atmospheric constituents such as methane.

We are developing a modeling framework to simulate high spectral resolution observations and retrievals using both FIR and MIR spectra, using an optimal estimation algorithm [C. Rodgers, 2000]. The retrieval algorithm is applied consistently on the simulated observations from both spectral regions. With this approach, we can objectively compare the available information content in the MIR and FIR.

We present results from our initial work exploring cloud – free atmospheric conditions. The climatological a priori information is generated from radiosonde data collected at the Atmospheric Radiation Measurement ground sites. These sites include midlatitude, arctic, and tropical climates, enabling comparison of the two spectral regimes over a wide range in temperature and moisture profiles. We use the LBLRTM line – by – line radiative transfer code for a highly accurate forward model [S. Clough, 2005]. In addition to the climatological variation, we introduce additional variation in the methane profile and the surface temperature. These represent unknown errors, which are unaccounted in the retrieved state estimate, and have different impact on observations in the two spectral ranges. The objective comparison is based on the degrees of freedom for signal, as well as the RMS error reduction of the retrieved profiles as compared to the climatological distribution.