A Simulation Study On The Methods Of Atmospheric Profiles Retrieval In Clear Sky With AIRS Observations

Atmospheric Infrared Sounder (AIRS) is the most advanced infrared atmospheric remote sensing instrument with super-high spectral resolution at present in the world. It has 2378 spectrum channels ranging from 650 cm^-1 (15μm) to 2700 cm^-1 (3.7μm) and a spectral resolution (v/△v) higher than 1200. Its ability of global overlay makes that it is viable to obverse the atmosphere and its variability on a global basis, further to provide large-scale and high-precise initial fields of humidity and temperature for numerical forecasting models. On this account, it is scientific and useful to study on processing and application of the super-high spectral resolution data.Global clear atmospheric train profiles data provided by University of Wisconsin are used here, half part as training sample, the other for retrieving and testing. Furthermore, AIRS simulating data derived from SARTA (Stand-Alone Radiative Transfer Algorithm) are used as regression factors to retrieve the atmosphere profile of temperature, humidity and other variables on a clear sky basis by means of eigenvector regression algorithms. The results are compared with observations to evaluate the retrieve. In order to improve the retrieve of the atmospheric temperature and humidity profiles, surface temperature, IR surface reflection index are introduced as restraint factor for retrieve equation. Analysis shows that:1. Eigenvector regression algorithms is a useful way to retrieve the atmospheric profile of temperature and humidity with speediness and better accuracy. The error of temperature profile is 3～4 K in middle and upper levels and 4～5 K below the troposphere. The average error of humidity profile is 1.23g/kg, and the max is 3.2g/kg locating on surface skin.2. Temperature profile retrieve is sensitive to characteristics of the earth's surface. The temperature in low level is evidently proved when the surface temperature is added, while that in 200-400mb is ameliorated when IR surface reflection introduced. If they are introduced at the same time, the profile in both low and upper level is better by 0.65/0.75K RMS (Root-mean-square error) averagely in low/upper level, which is remarkably better than that with single factor.3. The atmospheric humidity profile is not markedly improved when single restraint factor is introduced. Especially the effect of IR surface reflection is weak, while that of surface temperature is positive somewhere and negative other places. If they are added together, the profile near surface is nearly changeless, while it is relative evidently improve in 900-500mb. It is better by 0.4g/kg RMS in 900mb.