Study On Retrieval Method Of Ozone Profile From Satellite Limb Observation

Since the signing of the Montreal Protocol and its amendments,the related ozone research is still a hot issue in the field of atmospheric chemistry.The key to understand and predict the global environmental issues such as ozone depletion and climate change is an accurate global data set of atmospheric constituents with sufficient temporal and spatial(vertical and horizontal)resolution.Satellite measurments provide more comprehensive geographical coverage and regular observation.The limb sounding technique,widely used by more recent satellite instruments,combines the advantages of traditional nadir observation and occultation mode,that is,it has good vertical and horizontal resolution at the same time.The research of ozone profile retrieved for limb scattering radiance mainly uses wavelength pairing or differential optical absorption spectroscopy(DOAS)as the technology of spectral analysis and processing,and combines the optimal estimation algorithm to retrieve the ozone vertical profile.To improve the accuracy of atmospheric ozone profile,the retrieve of ozone profile based on the combination of wavelength pairing and weighted multiplicative algebraic reconstruction technology from the limb UV-VIS scattering radiation data are investigated in this paper.First of all,tangent height corrtion is an important prerequisite for obtaining high-quality ozone profiles from limb scattering observations.Aiming at the tangent height offset caused by the misalignment of limb detection instrument in the pointing process,a method based on UV narrow band tangent height retrieve is proposed.According to the characteristics of the knee of the limb radiation in the UV band,this method uses six wavelength(300-305 nm)radiation profiles to make the measurement vector not only maintain full sensitivity but also reduce the dimension,and combines the optimal estimation algorithm to retrieve the tangent height deviation.Compared with the traditional knee point method,the average accuracy of the algorithm is improved by 45%,while the running time of the algorithm is reduced by 20%.The average deviation of tangent height obtained is less than 150 m,which effectively reduces the ozone retrieval error caused by tangent height migration.Using wavelength pairing and weighted multiplicative algebraic reconstruction techniques,combined with the SCIATRAN forward model,the 10-69 km(cloudless)ozone number density profiles with a vertical resolution of 1 km are retrieved from the limb UV-VIS scattering radiation of the SCanning Imaging Absorption spectroMeter for Atmospheric Chatography(SCIAMACHY)carried on the ENVISAT satellite.Firstly,in order to reduce the influence of instrument calibration error,the limb radiation profile used for retrieval is normalized to the higher altitude(reference tangent height).Secondly,in order to maintain or improve the sensitivity of the retrieved components,the retrieved wavelengths are paired,and the wavelength pairing combination is optimized according to the instrument spectral channel and measurement technology,thus a new set of discrete wavelength spectra is proposed to form the retrieved vectors.Third,the impact of weight factors on retrieval in weighted multiplicative algebraic reconstruction technology is investigated,and a method based on the limb weighting function profile of the wavelength is proposed to calculate the wavelength weight factors to reduce the uncertainty caused by the empirical method.Finally,the simulated limb radiance model based on SCIATRAN model is established according to the limb observation geometry to retrieve the atmospheric ozone profile.The formal error analysis method based on weighted multiplicative algebraic reconstruction technique is used to analyze the main errors of ozone profile retrieval.In this method,numerical simulation is used to quantify the impact of various error sources on ozone retrieval,and the importance of each error source is characterized in detail.The maximum systematic error of ozone retrieval caused by main error sources is～17%,and the random error is about 5%.In addition,the limb radiance is greatly affected by the solar zenith angle,thus the sensitivity of each error source to the solar zenith angle is investigated.The error sources most affected by the solar zenith angle are aerosols,surface albedo and clouds.Finally,the error estimates of ozone retrieval between the northern and southern hemispheres under the same solar zenith angle are studied.These results are very useful for the strict verification of ozone profiles retrieved by different instruments,algorithms and models,and provide explanations for differences in data sets between instruments.To verify the ozone profiles obtained from limb scatterred radiance based on the retrieval scheme in this paper,the ozone retrievals are compared with other passive remote sensing satellite data and ozone sonde data.The consistency between the retrieved ozone profile and SCIAMACHY v3.5 provided by Bremen University is less than±5%within 20-38 km and controlled within 10%above 40 km,while the relative error of 10-15 km is relatively large(15-23%).Compared with OSIRIS v7.1,the average difference between 17-38 km is controlled within 10%,while there is a negative deviation of about 10-15%at 46-52 km.Compared with the ozone sonde data,the retrieved ozone concentration is about 3-5%higher than the measured value of the ozone sonde at an altitude of 10-35 km.In the correlation comparison of stratospheric ozone column,the correlation with SCIAMACHY v3.5 is the smallest,0.85,while the correlation with OSIRIS v7.1 and ozone sonde data is 0.86 and 0.89 respectively.Through the research in this paper,the limb tangent height correction is completed by using the UV narrow band and the optimal estimation algorithm.Using wavelength pairing and weighted multiple algebraic reconstruction technology,the retrieval of ozone profile from limb scattering radiation is realized,and the error of the retrieval result is analyzed in detail.At the same time,the validity and reliability of the retrieval method in this paper are verified by comparing with other data sets of the same period.