Cloud Detection Method And Application Research For Hemispheric Scale Data

Clouds cover most areas over the earth and play an important role in the Earth’s water circulation,energy balance of the earth-atmosphere system and radiation transmission.At the same time,clouds are the most common meteorological phenomena.The amount of clouds and the distribution of clouds directly affect the weather,the weather change,and have an important significance for human daily life,industry and agriculture.In addition,the cloud coverage of ground surface information leads to many problems in the image registration,fusion and other processing processes.Cloud detection is also regarded as the key to image recognition,classification and interpretation.So cloud detection is very important.This paper studies the Earth’s Polychromatic imaging camera(EPIC)data on the Deep Space Climate Observatory(DSCOVR)satellite launched in 2015.Based on the wide range of EPIC data and large coverage(hemispheric scale),one new cloud detection method for the hemispherical scale data is proposed.First,we analyze the different reflection characteristics of different EPIC bands,which are determined by the physical properties of objects.In particular,the ultraviolet bands of EPIC data are new.Then,based on the cloud-sensitive bands in the EPIC data,the selected bands are combined in the form of an index to obtain a new cloud index.Based on the defined new cloud index,the cloud detection of the EPIC hemispheric scale data is completed.To verify the effectiveness of the proposed hemispherical scale cloud detection method,the EPIC data that correspond to summer(July 3,2017)and winter(January 3,2017)are used to conduct the experiments.Three other commonly used cloud detection methods,namely,the visible light cloud detection method,support vector machine(SVM)cloud detection method,and normalized cloud index(NDCI)cloud detection method are compared.The comparison results show that the proposed cloud index cloud detection method is superior to the visible light cloud detection method,the SVM cloud detection method and the normalized cloud index cloud detection method in terms of the cloud distribution and cloud amount.The cloud distribution obtained by the proposed method is most consistent with the cloud distribution in the original EPIC image combined with RGB true color.The correct detection ratio,missed detection ratio,false detection ratio and Kappa coefficient are used to evaluate the accuracy assessment.The method is also verified by the EPIC L2 data.It is shown that the proposed method is valid and reliable,and can quickly obtain cloud distribution and cloud amount within the hemisphere,which is helpful for dynamic research and natural weather prediction of global clouds.In order to support the dynamic research of clouds,the relationship between clouds and time is studied based on the cloud detection.The relationship between the cloud area and the latitude and longitude position,the cloud frequency within one day,and the cloud frequency of the first,middle and last ten-day period of a month are analyzed.The extended application research can be used to study the global cloud distribution characteristics and the dynamic research of clouds in a day or more days,which will be contributed to the prediction and analysis of the cloud movement trend.