Research On Radiative Transfer Characteristics And Retrieval Methods For Atmospheric CO₂Monitoring

Atmospheric CO2is one of the most important Green House Gases. The sources and sinks of CO2could have important influence on climate change and human activities. But the concentration of CO2is very low and the variation is quite small in atmosphere, which made CO2monitoring difficult. Compared to ground sites, spaceborne observations could provide better spatial coverage which made it possible to monitor CO2concentration globally. As a major carbon emitting country, China has started its own spaceborne CO2observation projects. Thus research on radiative transfer characteristics and retrieval method for atmospheric CO2remote sensing became necessary to fullfill the demands of economic development and environmental diplomacy. Firstly, the radiative transfer characteristics of CO2were discussed. Then retrieval methods for CO2column and profile as well as some techniques to improve retrieval performances were developed. At last, retrieval methods were validated by satellite observations.Atmospheric molecules and particles may affect the energy passing through them. Precise description of such processes is the foundation of atmospheric remote sensing. According to basic theories of radiative transfer, the interferences which may influence CO2absorption signitures were simulated in different CO2absorption bands. Simulation results suggested that1.6um band in near infrared is more suitable for spaceborne CO2retrieval because of weak absorption and less interference by other gases. Then the reason for spectral resolution degeneration in observation was discussed with a Fourier Transform Spectrometer(FTS) model, and the sensitivity analyst was done under different resolutions using this model. Furthermore, CO2weighting functions in near infrared (NIR) and thermal infrared (TIR) were also calculated to demonstrate sensitivity differences on different altitudes.Based on the research of radiative transfer characteristics, retrieval methods and improvements on calculation speed and precision were developed. Firstly, a three dimensional look up table (LUT) was established to avoid the complicated line-by-line calculation. In the LUT, gas absorption cross sections were expressed as a function of wavenumber, temperature and pressure. Extracting gas absorption cross sections from such LUTs could be much faster than direct line-by-line calculations. Besides, LUTs made it possible to update weighting functions after each iteration in retrieval. The LUT based forward model is also easier to be integrated with retrieval algorithms. Using such LUT based forward model, a retrieval method based on differential optical absorption spectroscopy (DOAS) was demonstrated. Then maximum a posteriori contrain was introduced to profile retrieval to ensure the validity of retrieval results. Furthermore, it is shown that retrieval results from high resolution spectra could easily be influenced by wavelength distortion, and an algorithm using nonlinear optimization was established to correct such effects.In order to validate the DOAS-based retrieval method, CO2concentration from Park Falls ground site were selected as reference. And the spectra collected by GOSAT satellite over Park Falls were analysed using the DOAS-based retrieval method. Retrieval results were compared with CO2concentrations obtained by Park Falls ground site as well as corresponding GOSAT Level2products. All three datasets showed same tendency of CO2concentration change. At last, the profile retrieval method using maximum a posteriori constrain was validated by simulated retrieval due to lack of in-situ CO2profile measurements.Profile retrieval using both NIR and TIR showed that combining both bands could reconstruct CO2verticle distribution better than using NIR or TIR band alone, which suggests a potential to further improve retrieval precision.At last, the DOAS-based retrieval method were slightly modified so it could be used to analyse ground observed high resolution spectra. The retrieval results were compared with concentration data provided by Max Planck Institute for Biogeochemistry, and they fit well. Which demonstrated the method’s potential to validate satellite retrieval results by using ground observations.