| Using MODIS remote sensing of sea fog is an effective technique to study the sea fog. But due to the lack of marine observation data, it is difficult to obtain more precise data to do further research about MODIS-based detecting sea fog. Spaceborne lidar CALIOP can make the three dimensional observations of all kinds of atmospheric targets in a global range and it has a high vertical resolution data structure with 30 m below 8.2 km. Firstly this paper carries out the research of a method of sea fog discrimination by using CALIOP data and presents a sea fog detection method of combining level-2 VFM product and level-1 backscattering coefficient data. Then the global distributions of sea fogs and clouds, including the occurrence probabilities and the statistics of thickness, are studied based on this method. At the same time large amounts of sea fog, low cloud, middle and high cloud, clear air samples are extracted and used in the studies of channels selecting and static thresholds setting for MODIS daytime sea fog detection.Firstly there are lots of misjudged sea surface samples in the CALIOP level-2 Vertical Feature Mask product VFM. Combined with the 30 m-resolution altitude information, misjudged samples whose top heights are greater than 60 m are extracted. Then this paper has analyzed the difference of scattering coefficient between low clouds and aerosols through the statistics of 35 sea fog cases. Results shows the scattering coefficient of most aerosols is less than 0.03 km-1sr-1 in 532 nm and less than 0.025 km-1sr-1 in 1064 nm. Aerosols among the misjudged samples are deleted and clouds next to sea surface which are fogs are extracted with these two values as threshold. Compared ICOADS data of 100 years with CALIOP data of 2008 about the sea fog occurrence frequency on North Pacific in January, April, July and October, result shows this method is feasible.In this paper, the global statistical distribution of sea fog occurrence probability, sea fog thickness, low cloud、middle and high cloud occurrence probability, cloud layer numbers occurrence probability and cloud thickness is calculated based on 2011 CAL1OP level-2 VFM and CLay products. These achievements can not only prove the data accuracy of CAL1OP level-2 products and the applicability of sea fog research on East China Sea. but also can provide an important reference for the following research and analysis.Matching the geographic information of sea fog samples, low cloud samples, middle and high cloud samples and clear air samples extracted by CAL1OP with Aqua-MODIS reflectance and radiance bands data, channels are selected and thresholds are set on the principle of that fog samples are as much as possible in the multi-channel thresholds while the other atmospheric target sample are as little as possible in the multi-channel thresholds. Among the reflectance bands,6th and 9th-16th channels have a large amount of invalid data in the cloud zone which account for more than 40% of all data. So these channels used for ocean color, biochemistry and other purpose are ignored in this paper. MOD1S 26th channel is dedicated to the distinguishment of cirrus clouds, so thresholds 0.06 is set to eliminate the cirrus cloud. Through man-machine interaction thresholds are set in the remainder 10 reflectance channels. First step is to find the channel that has the highest accuracy rate on sea fog discrimination, which results to be 8th channel. Then in the assurance of sea fog omissive judgement rate less than 40%, all remainder channels are fused with 8th channel one by one. Through the contrast of accuracy rate,7th channel and 8th channel are found to be the most suitable channels to identify sea fog among all reflectance channels. Applying this method to radiance bands, adding the threshold limits of 32th channel is the most effective way to improve the sea fog detecting efficiency. Finally two sea fog cases on East China Sea are detected using these static thresholds, and through comparison of two sea fog cases on East China Sea with ground stations data and ship records the detecting results are verified. |
PDF Full Text Request