Research On Infrared Hyper-spectral Clear Channnel Detection Used In Variational Assimilation Method

The development of real-time and refinement in numerical weather forecast technology requires new requirement to atmosphere observation. As a kind of atmospheric vertical sounder, The Atmospheric Infrared Sounder(AIRS) can provide global observations four times per day, which has characters of high space resolution, high spectral resolution, and high radiometric resolution.AIRS has 2378 channels which will bring huge compute pressure to direct assimilation system. High correlation of each channel can also bring bad effect to the assimilation result. For these reasons, the radiative transfer model which fit for infrared hyper-spectral data also develop some new method. This paper starts with hyper-spectral radiative transfer equation, studies the character of AIRS weigthing function. In the design of infrared hyper-spectral assimilation observational operator's spatial convert scheme, because of the weighting function of AIRS distribute very close and the sharpness is very pointed, the linear vertical interpolation may cause oscillation in Jacobian profile mapping to different vertical coordinate. The paper brings the Piecewise log-linear averaging interpolation scheme, improve the character of Jacobian profile in different vertical coordinate mapping.Cloud can be treat as blankbody in infrared radiation, and the clear FOV which is not polluted by cloud only takes 10 percent of all observed FOV, it will be a huge waste of data if we cursory discard the polluted FOV data directly. Cloud detection is the indispensable key step in the quality control of infrared hyper-spectral data, The clear channel cloud detection scheme can distinguish clear channel, so this scheme can avoid data waste and it can bring some cloud-top information to assimilation system. WRFDA system contain NESDIS-Goldberg cloud detect scheme which based on clear FOV theory. In this peper, we discuss the principles of every part of clear channel cloud detection scheme and achieve this scheme in the base of WRFDA system. We design a new parameter to solve the low accuracy of channel height. And we also fix the wrong clear FOV judgment in low-cold situation in clear channel search algorithm.We compare with two kinds of cloud detect scheme in cloud detect effection, analysis field, forecast field, tyhoon track based on assimilation-forecast experiment system, the result shows:(1) The amount of upper-atmospheric channels(chan1766,6600)which are used to assimilate is more than lower atmospheric channels(chan362,5800) with the scheme of clear channel cloud detection;(2)The capability of clear channel cloud detect scheme is stronger than NESDIS-Goldberg scheme;(3)Analysis field which comes from clear channel cloud detect scheme has smaller RMSE.(4)Clear channel cloud detect scheme can bring more troposphere-cloud-top radiation information to assimilation system, it can obviously improve accuracy of temperature, humidity at lower troposphere level and wind field at higher troposphere level;(5)Typhoon track which comes from clear channel scheme's forecast field is more similar to actual observation.