| The 17 th World Meteorological Congress,held in 2015,identified the 2040 Development Plan for the WMO Integrated Global Observing System?WIGOS?,led by the World Meteorological Organization?WMO?Basic Systems Committee?CBS?.Each WMO's requirement for observing systems is higher,faster and more accurate.The future global numerical weather forecast will provide more comprehensive and accurate forecast information for mankind,and needs more atmospheric variables provided by meteorological satellites,thus three-dimensional atmospheric temperature and humidity distribution data of sufficient vertical resolution is very important.Fourier Transform Spectrometer,compared to other types of hyperspectral instruments,offers high luminous flux,multichannel and hyperspectral resolution,making it ideal for atmospheric detection.Since 1962,the Fourier Transform Spectrometer has been continuously carried into the spacecraft for observation of atmospheric temperature and humidity,water vapor height distribution,process of atmospheric transformation,landmark components and atmospheric pressure.The FY-4 meteorological satellite was launched on December 11,2016,which is the second-generation geostationary orbit?GEO?remote sensing meteorological satellite in China.The three-axis stabilization control scheme is used to replace the spinstable satellite?FY-2?.And this satellite greatly improves the capacity of remote sensing in geostationary orbit.The FY-4 satellite is equipped with Geostationary Interferometric Infared Sounder?GIIRS?with a spectral resolution of 0.625 cm-1,enabling highprecision detection of the atmospheric structure in the vertical direction.GIIRS is the world's first successfully operating interferometer in GEO,even the United States thirdgeneration geostationary satellite GOES-R is not equipped with it.In this paper,the research of infrared Fourier spectrometer information processing technology,mainly concentrated in three aspects: first,the Fourier spectroscopy model research and analysis,including the study of sampling and instrument line shape;the second is to study the spectral inverse algorithm,through the evaluation and correction of the interferogram,proposing the method to improve the accuracy of inversion spectral;third,the high-speed implementation of spectral inversion algorithm,using high-performance hardware platform to achieve rapid inversion of the spectrum,which could solve the real-time problem.Before the paper begins,the main theory of Fourier spectrometer is introduced.The noise of Fourier spectrometer is analyzed.And the formula of SNR of Fourier spectrometer is deduced.The theoretical method of improving the SNR of instrument is obtained.This paper analyzes the sampling of Fourier Transform Spectrometer,analyzing the influence of equal time interval sampling and equal optical path difference?OPD?sampling on the spectrum.This paper also researchs the digital resampling method,which is proposed by Brault ans so on.According to the interpolation method,the digital method resampling can be divided into three kinds: polynomial interpolation resampling,Fourier transform interpolation resampling and Brault resampling.This paper introduces the method of simulating interferogram,and realizes the verification and analysis of theory by using the simulated interferogram.This paper systematically analyzes the resampling method by using the simulated interferograms,and gives the simulation results.The results show that using the polynomial interpolation resampling and Brault resampling can reduce the RMS noise of the original normalized spectrum from 10-3 order to 10-16 and 10-5.Brault resampling compared with the other two interpolation methods,has a high computational efficiency with good reduction in sampling noise.Due to the complex spatial environment or the inherent characteristics of spaceborne Fourier Transform Spectrometer,the obtained spectrum has additional noise or spectral offsets compared to the real one.When dealing with interferometric data,these factors need to be taken into account.And we should take measures to eliminate or reduce the impact of these factors on the final spectrum.In this paper,the main algorithms of data process for FTS,such as spikes detection,nonlinear correction,apodization,bandpass sampling,Fourier transform,spectral calibration and radiometric calibration,are studied theoretically.And this paper also uses simulated data and laboratory calibration data to verify and analyze them.In the study of spectral calibration,the Fourier Transform Spectrometer is quantitatively analyzed,and the Instrument Line Shape?ILS?is obtained.Using the instrument linear function,the wave number error between theory value and the actual value is only 4.21%,which are the values calculating the wave number difference between the central pixel and the edge pixel in the detector array.After correction by the theoretical difference,the spectral accuracy reaches 8.38×10-5.Based on the current device and technology level,this paper proposes an information processing scheme from the original interferogram to spectrum for the spaceborne Fourier spectrometer.The scheme has made many improvements to the existing excellent algorithms and proposes some innovative algorithms.Efficient engineering application is also taken into consideration.In this paper,the non-linearity of the detector,ZPD position detection,the use of filtering and decimation for data compression,as well as parallel CZT calculation are innovatively researched.Moreover,the proposed algorithms are verified and analysed by using the experimental data or simulated data,the effect is excellent.An algorithm for calculating the nonlinear coefficient of the photoconductive infrared detector is proposed based on the experimental calibration data.The experimental results show that the nonlinear quadratic coefficient of the long wave interferogram is 10-6 order,and the result is valid.The data compression is carried out by filtering-decimation and bit reduction.Through this method,the spectral SNR keep same,but the minimum amount of longwave data is only 10.9% of the original data,and the minimum amount of medium wave data is 14.5%.Finally,this paper introduces the real-time spectral inversion technology using multi-core DSP hardware platform.The choice of multi-core DSP can not only meet the needs of the current system,but also has a reference for the future high-speed implementation of spatial data processing.In this paper,the main technologies involved in multi-core DSP,PCIe driver development and lower computer software development were introduced,and the platform computing performance was studied and analyzed.The algorithms of parallel FFT and CZT is verified by this platform,and operating results are good.When operating on 32 K float point data,the parallel FFT operation uses 2 ms,and the absolute error is 10-4 order compared to the 64-bit float point calculation by MATLAB.The parallel CZT operation costs 2 ms,and the absolute error is 10-4 order.In this paper,the information processing of the spaceborne Fourier Transform Spectrometer is studied systematically,and a complete solution is proposed for the more accurate and faster spectral restoration of the spaceborne Fourier Transform Spectrometer,which is of great reference for the future space application of Fourier Transform Spectrometer. |
PDF Full Text Request