Sensitivity Analysis And Algorithm Improvement Of Atmospheric Carbon Dioxide Retrieval Using Satellite Remote Sensing

Global real time carbon dioxide data with high accuracy is of great significance for carbon cycle and global change study, and using satellite remote sensing technology and radiative transfer model to retrieve atmospheric carbon dioxide has become the most important means to obtain the global carbon dioxide concentration data. Supported by the National Basic Research Program of China, this paper made the following research and improvement of the current carbon dioxide retrieval process.Firstly, considering the uncertainty of profile data, a perturbation method based on root mean square error was designed to do the sensitivity analysis, and the sensitivity of temperature profile, vapor profile, ozone profile, surface temperature and emissivity were compared with that of carbon dioxide. Result showed that temperature profile was the main factor that affects the carbon dioxide retrieval accuracy. Then the optimal carbon dioxide retrieval channels were selected based on SNR (signal to noise ratio).Secondly, hyper-spectral remote sensing data has a lot of channels with too much information, which makes it very time-consuming for researchers to analyze the characters of different parameters on each channel. A Hyper-spectral Channel Character Visualization and Analysis System (HCCVAS_V1.0) was designed and implemented for the visualization and comparison of sensitivity data, SNR data as well as jacobian data, so as to improve the research efficiency.Finally, as the temperature profile error would greatly affect the carbon dioxide retrieval accuracy, a Temperature Error Elimination method (TEE) based on channel combination was designed by improving the regular retrieval cost function to reduce the influence of temperature profile error. Then this method was programmatic implemented on Linux platform and the effectiveness was further verified. It turned out that TEE did reduce the influence of temperature profile error and the carbon dioxide retrieval accuracy with TEE was better than the regular method at areas of low latitude and high temperature.Under the background of global warming, this research studied the key steps of the carbon dioxide retrieval using hyper-spectral remote sensing data, in order to provide some reference for the further study on quantitative carbon dioxide measurements.