Research On The Performance Estimation Of Space-borne High-spectral-resolution Lidar And Its Denoising Algorithm

According to the IPCC(The Intergovernmental Panel on Climate Change)report,of the many factors affecting the Earth's climate change,the impact of aerosols and clouds and their interactions on climate change attributes the greatest uncertainty.On the other hand,the atmospheric environment is the environment in which humans are in direct contact.Pollution in the atmosphere has a direct and irreversible impact on human life and health.As an active remote sensing instrument,lidar has the advantage of high spatial and temporal resolution and can work both at day and night.Compared with other observation instruments,it is irreplaceable.Furthermore,the space-borne lidar can measure the global cloud aerosol profile distribution thanks to its wide coverage,and it has become a key tool for scientists to understand the changes in the earth's atmosphere.Since launched in 2006,Cloud-Aerosol Lidar with Orthogonal Polarization(CALIOP),which is still on orbit,has provided more than a decade of data.China is developing its own Aerosol-Cloud High-Spectral-Resolution Lidar(ACHSRL),which uses an iodine molecular absorption cell as a frequency discriminator,and this satellite is about to launch.An assessment of detection performance is needed before launch.On the other hand,for space-borne lidar,it is important to study effective denoising methods for space-borne lidar.The main research contents of this thesis are as followsAn error model for inversion of optical characteristics of space-borne high-spectral-resolution lidar is established.Through the simulation with system parameters of ACHSRL,the detection capability of ACHSRL and the retrieval accuracy of its optical properties of cloud aerosol particles in the atmosphere are analyzed for the first time.In addition,the historical data of CALIOP was used as the input of the atmospheric model to evaluate the inversion accuracy of global land measurement in summer by ACHSRL.Furthermore,we compared the retrieval accuracy of the ACHSRL and CALIOP in the aerosol transfer area.The error analysis model indicates that,at the same resolution as the CALIOP Level 2 Profile product,73.63%of the backscatter coefficient relative error for ACHSRL is less than 40%,and this index is 30.72%for CALIOP.As for extinction coefficient absolute error,76.01 percent of ACHSRL is less than 0.2 km-1,and this number of CALIOP is 56.97%.The analysis shows that the inversion accuracy of the backscattering coefficient for ACHSRL is significantly improved compared to CALIOP,and the inversion accuracy of the extinction coefficient is also refined to a certain extent.In order to further improve the inversion accuracy of ACHSRL and overcome the influence of noise,we have further studied the space-borne Lidar denoising algorithm suitable for ACHSRL,which will fly in space at high-speed.Compared with the traditional smoothing or single profile fitting algorithm,this paper utilizes BM3D(Block-matching and 3D filtering)algorithm,which is the state-of-art in the field of image processing.Since ACHSRL has not yet been launched,a simulation study of ACHSRL is performed based on the historical data of CALIOP.We connect many adjacent lidar signal profiles and treat them as images.Unlike existing lidar noise reduction algorithms that use unselected adjacent bins for smoothing,BM3D performs a frequency domain transformation on the signal image and then searches for similar blocks in the frequency domain for collaborative filtering.This algorithm not only achieves good denoising effects,but also retains aerosol/cloud feature edge details.After BM3D denoising,the peak signal-to-noise ratio(PSNR)of the echo signal in all channels are improved,and the inversion accuracy of ACHSRL is also improved,especially for extinction coefficient.In addition,we applied the BM3D algorithm to the deep learning network layer detection of CALIOP,which enabled the layer detection for daytime low signal-to-noise ratio CALIOP data.