Research On Atmospheric Correction Of High Resolution Remote Sensing Imagery In Urban Areas

Due to the existence of a large number of man-made targets in urban areas,such as roads,buildings and vegetation,the surface changes are drastic and complex,and the existing surface reflectance products cannot be effectively monitored at fine-grained levels.Atmospheric correction can remove the atmospheric effects of remote sensing images.To restore the clarity of the image and the real surface reflectance,atmospheric correction of high-resolution remote sensing images is beneficial to quantitative remote sensing research in urban areas,and is of great significance.This thesis uses Sentinel-2 high-resolution remote sensing images to conduct atmospheric correction research,aiming to remove the influence of atmospheric components in remote sensing images and obtain the surface reflectance of high-resolution remote sensing images in urban areas.Quantitative atmospheric correction needs to solve two problems,one is the acquisition of atmospheric parameters,and the other is the quantitative correction of atmospheric parameters.In this thesis,the retrieval of atmospheric aerosol parameters is proposed using the ratio relationship between the red and coastal blue bands of Sentinel-2,and the subsequent atmospheric correction process is improved on this basis.The main work includes the following aspects:(1)To study the influence of atmospheric parameters on the radiative transfer,based on the characteristics of the Sentinel-2 sensor,the two-band algorithm was selected to retrieval the water vapor information,and the water vapor information from AERONET ground monitoring was used to verify the accuracy of the water vapor retrieval.Sexuality R2 reached 0.962.At the same time,the expected error also reaches 68%.The test in the Beijing experimental area shows that the retrieval results have good spatial distribution characteristics,which is helpful to remove the water vapor absorption effect of the subsequent atmospheric correction and improve the accuracy of atmospheric correction.(2)Considering the complex action mechanism of aerosol on visible light and the relatively stable surface reflectance ratio relationship in urban areas,an improved aerosol retrieval algorithm that does not require pre-estimation of surface reflectance is proposed.This method is based on the monthly ratios factor map synthesized from the surface reflectance products of MODIS,the radiative transfer model and the look-up table method to perform aerosol retrieval.The experimental correlation R of the retrieval experiment in Beijing area reaches 0.91,and the inversion results are 69.71%fall.With in the expected error.It is proved that the aerosol retrieval results in the urban area of Beijing are reliable,which is beneficial to the effective removal of subsequent aerosol scattering effects.(3)Based on the above retrieved atmospheric parameters and the principle of radiative transfer,the improvement of the atmospheric correction process is realized,and the quantitative atmospheric effect removal experimental research is carried out on the Beijing area on the basis of the cirrus cloud correction,and the atmospheric correction results from Analyzed from both qualitative and quantitative perspectives.The results show that the atmospheric correction algorithm in this thesis can effectively improve the contrast and clarity of the image;the correlation with the products produced by the official Sen2Cor algorithm of Sentinel-2 is above 0.95 except for red light,and the verification of vegetation monitoring ability in urban areas also shows good results.In addition to the EVI index that is affected by blue light,the correlations of other index indicators are all above 0.85.In terms of quantitative retrieval of aerosols,the algorithm in this thesis has significant advantages,while the results of the Sen2Cor algorithm are high and the error is large.However,both of them have a correlation of more than 0.816;compared with the surface reflectance data of the ground truth experiment,it shows a strong consistency.Except for the 945 nm water vapor absorption band,the correlation coefficient R reaches 0.9458.