| The formation and evolution of cloud is closely related to meteorological climate change.The results of continuous observation of cloud can be widely used in aviation,navigation,military,meteorological analysis,solar energy capacity forecasting and agricultural production.Therefore,it is of great significance to observe cloud parameters.Ground-based cloud observation is one of the important ways of cloud observation.In the early days,manual visual observation is the main method,but it cannot provide continuous,objective and quantitative cloud parameters.For the needs of ground-based cloud observation,automated ground-based cloud observation equipment based on different principles has been proposed and developed.Among them,the all-sky imaging equipment has the advantages of high resolution of local spatial and temporal observation,low influenced by meteorological conditions and continuous day and night observation,etc.It is widely concerned.However,there are still shortcomings in the structural design of existing all-sky imaging devices,and the accuracy of cloud parameter inversion algorithms and their applicability under different weather conditions are limited due to the complex and variable cloud structure and morphology,so far they cannot really meet the demand for automated ground-based cloud observation.In this paper,according to the status of ground-based cloud observation research,a dual-band all-sky cloud imager is proposed and developed.It is designed to image in the visible band(450～650nm)and infrared band(8～14μm)through a dual-channel optical path,and works in the visible plus infrared observation mode for all-day groundbased cloud observation.Its visible imaging unit,through the built-in chopping plate design,and combined with high dynamic range imaging technology,can collect clear,unobstructed visible cloud images of the whole sky;its infrared imaging unit,different from the traditional time-for-space scanning stitching design of observation,using a large array infrared detector with a custom infrared wide-angle lens,to achieve a single field of view of the sky infrared imaging not less than 160°.It is the largest single observation field of view among similar equipment at present.To address the observation needs of macroscopic parameters of cloud,this paper carries out research on the inversion algorithm of ground-based cloud parameters based on the observation data of dual-band all-sky cloud imager.To invert cloud parameters,deep learning techniques are introduced into the segmentation task of visible cloud image,and a cloud image segmentation network SegCloud is proposed to overcome the shortcomings of traditional cloud image segmentation methods based on texture,greyscale and other low-order visual information which are greatly affected by weather conditions,and can effectively segment visible cloud image.In the infrared band,an infrared cloud image segmentation algorithm based on the numerical simulation of infrared raw grey-scale image is proposed,giving the all-sky infrared cloud image background radiation correction formula and an infrared cloud image segmentation method.In order to verify the effectiveness of the cloud image segmentation algorithm,this paper analyses the data in terms of both accuracy and consistency.Cloud cover information is calculated from the cloud image segmentation results,and the visible and infrared cloud cover are compared with manually observed cloud cover.The comparison results effectively demonstrate the feasibility of the dual-band cloud image segmentation algorithm.In addition,ground-based cloud base height inversion algorithm is developed based on the infrared band all-sky bright temperature data.Combining the decreasing atmospheric temperature characteristics and laser cloud height meter data,the decreasing atmospheric temperature coefficient of the cloud base is calibrated and the inversion formula of the zenith cloud base height is derived,and then the cloud base height within the 120° observation field of view of the infrared sensor is derived from the zenith cloud base height.This paper systematically carries out the research and development of dual-band all-sky cloud imager and cloud parameter inversion algorithm,overcoming many problems of existing all-sky imager equipment in imaging,realising the acquisition of all-sky visible and infrared cloud images in a single pass,and achieving continuous observation of cloud day and night.The simple structure and low cost of the equipment are of high application value.Based on the observation data of this equipment,not only the macroscopic parameters of ground-based cloud can be inferred,providing a new means of automated observation of ground-based cloud,but also the inversion of parameters such as optical thickness of cloud and precipitable water,as well as the fusion with meteorological satellite data and the validation to obtain multi-channel raw ground-based cloud image data. |
