Research On Pollution Cloud Recognition Algorithm Based On POLDER Data

Clouds cover about 67%of the Earth surface.Through their multiple interactions with radiations,clouds have an important impact on the Earth-Atmosphere radiation balance and consequently on the climate.Large number of fine particles emitted from human activities.These small particles enter the lower atmosphere later pollute lower liquid cloud.At present,the effects of pollution on cloud which considered one of the main sources of uncertainty in climate change predictions.The precipitation formed in polluted cloud will greatly impact on human health and ecological environment.Identifying polluted clouds is of great significance.Since the polluted clouds contain many small partials,and polarized reflectance is sensitive to them.Therefore,the polarized sensing may be an effective mean to identify polluted clouds.The POLDER data with multiple angles and multiple spectra was used in this study.GF-5 satellite equipped atmospheric aerosol multi-angle polarization probe?DPC?which has the same polarization channel as POLDER.Studying polluted clouds by using POLDER data can lay the foundation for China's GF-5 satellite.This paper aims at identify polluted clouds based on multi-spectral and multi-angle polarized data from POLDER.Then the polluted clouds are furtherly divided into smoky pollution clouds and dusty pollution clouds.Through comparing cloud particle spectrum distributions for clean cloud and polluted cloud,we analyzed and simulated the differences between clean clouds and polluted clouds in polarized radiance at 865nm by using VLIDORT model.Then two thresholds,the decline rate and the maximum of the₉8))at 865nm in scattering angle ranging from 142°to 165°,have been designed to better identify clean clouds and polluted clouds.Dusty polluted clouds and smoky polluted clouds are typical natural pollution clouds and anthropogenic pollution clouds.The identification methods of the two clouds are constructed in this paper.The reflectivity and polarization reflectance of smoky and dusty pollution clouds were calculated in different bands,and the results were compared with POLDER data.Smoky polluted clouds tend to a strong positive reflectance gradient in the VIS to NIR spectrum.The fitting slope of the spectral reflectance is positive and this rising gradient is positively correlated with the degree of contamination.A larger polarized reflectance values at scattering angle range from 80°to 120°and smaller polarized reflectance values in the primary rainbow peak.Therefore,identifying smoky polluted clouds can use the reflectivity and the polarization reflectance.Dusty polluted cloud is a common natural polluted cloud.The reflectance of dusty polluted clouds is decreasing with increase of wavelength can be found.The polarization reflectance of dust pollution cloud increases very little at the scattering angle range from 80°to 120°.In addition,the polarization reflectivity peak at scattering angle 140°is reduced.These features can be used to identify contaminated clouds and dusty polluted clouds well.Finally,the algorithm was applied to different contaminated areas.The CALIPSO and AQI was used to valid the correctness of polluted clouds.The results showed that the algorithm can identify pollution clouds and accurately determine the types.This research can not only make up for the lack of algorithm of recognition polluted cloud,but also point out the type of pollution clouds.That lays a foundation for study of polluted clouds with China's GF-5 satellite.This helps to understand and quantify the effects of anthropogenic aerosols and natural aerosols on the cloud.Furtherly,the identification of contaminated cloud could assist in confirming the origin and types of pollutants,which is a new way to study the pollutions.