Day and Night Clouds Detection Using a Thermal-Infrared All-Sky-View Camera

Wang, Yiren; Liu, Dong; Xie, Wanyi; Yang, Ming–Hsuan; Gao, Zhenyu; Ling, Xinfeng; Huang, Yong; Li, Congcong; Liu, Yong; Xia, Yingwei

doi:10.3390/rs13091852

Cited by 16 publications

(6 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…3e. An ensemble of the results of each decision tree is obtained, and hence a prediction result is provided (Wright and Ziegler, 2017). That is, in the RF ensemble learning method, every individual tree of the decision tree contributes to the final prediction (Shin et al, 2020;Kim et al, 2021).…”

Section: Random Forest (Rf)mentioning

confidence: 99%

Twenty-four-hour cloud cover calculation using a ground-based imager with machine learning

2021

View full text Add to dashboard Cite

Abstract. In this study, image data features and machine learning methods were used to calculate 24 h continuous cloud cover from image data obtained by a camera-based imager on the ground. The image data features were the time (Julian day and hour), solar zenith angle, and statistical characteristics of the red–blue ratio, blue–red difference, and luminance. These features were determined from the red, green, and blue brightness of images subjected to a pre-processing process involving masking removal and distortion correction. The collected image data were divided into training, validation, and test sets and were used to optimize and evaluate the accuracy of each machine learning method. The cloud cover calculated by each machine learning method was verified with human-eye observation data from a manned observatory. Supervised machine learning models suitable for nowcasting, namely, support vector regression, random forest, gradient boosting machine, k-nearest neighbor, artificial neural network, and multiple linear regression methods, were employed and their results were compared. The best learning results were obtained by the support vector regression model, which had an accuracy, recall, and precision of 0.94, 0.70, and 0.76, respectively. Further, bias, root mean square error, and correlation coefficient values of 0.04 tenths, 1.45 tenths, and 0.93, respectively, were obtained for the cloud cover calculated using the test set. When the difference between the calculated and observed cloud cover was allowed to range between 0, 1, and 2 tenths, high agreements of approximately 42 %, 79 %, and 91 %, respectively, were obtained. The proposed system involving a ground-based imager and machine learning methods is expected to be suitable for application as an automated system to replace human-eye observations.

show abstract

Section: Random Forest (Rf)mentioning

confidence: 99%

Twenty-four-hour cloud cover calculation using a ground-based imager with machine learning

2021

View full text Add to dashboard Cite

show abstract

“…Several studies use IR or thermal-infrared (TIR) cameras to detect nighttime clouds (Aebi et al, 2018;Redman et al, 2018;Smith & Toumi, 2008;Y. Wang et al, 2021).…”

Section: Detection Of Nighttime Clouds and CC Algorithmmentioning

confidence: 99%

Observations of Nighttime Clouds Over Chiba, Japan, Using Digital Cameras and Satellite Images

2021

View full text Add to dashboard Cite

Nighttime cloud cover (CC) is analyzed from the data observed for 34 months over Chiba, Japan, using continuously operated digital cameras. The normal camera works in the visible region, providing quantitative and temporal changes of nighttime CC. The second camera captures sky images without the near‐infrared cut filter, and by applying the thresholding method, we obtain information on the temporal changes of thin (pixel values between 72 and 115) and thick (>115) clouds. Comparing the CC from the visible camera with cloud fraction from Atmospheric Infrared Sounder (AIRS) shows that long‐term data from the camera are valuable for correcting the underestimation of the cloud fraction product from AIRS for low‐level clouds. Furthermore, a comparison between the near‐ground temperature and CC results in estimating the seasonal changes of feedback effects of upward thermal radiation from the ground. The current results reveal that thick clouds contribute to longwave cloud forcing (LWCF) below 100 Wm−2, while thin clouds with CC below 30% contribute to LWCF below 10.2 Wm−2.

show abstract

“…The prevalent optical ground-based imaging equipment used for this purpose is a camera with fisheye lenses (Qi et al 2014). As for infrared, cloud bolometric brightness, ascertained through infrared radiation intensity, facilitates the retrieval of precise cloud temperatures and thicknesses, thereby providing valuable cloud distribution information (Liang et al 2017;Wang et al 2021;Wu et al 2022). Nonetheless, compared with visible light cloud measurement devices, infrared-based systems are often more expensive and demand frequent maintenance.…”

Section: Introductionmentioning

confidence: 99%

Cloud Identification and Reconstruction from All-sky Camera Images Based on Star Photometry Estimation

Zhi 支,

Wang 王,

Zhang 张

et al. 2024

PASP

View full text Add to dashboard Cite

Cloud cover significantly influences ground-based optical astronomical observations, with nighttime astronomy often relying on visible light all-sky cameras for cloud detection. However, existing algorithms for processing all-sky cloud images typically require extensive manual intervention, posing challenges in identifying clouds with pronounced extinction characteristics. Furthermore, there is a lack of effective means for detailed visualization of cloud cover. To address these issues, this paper proposes a method that reconstructs the cloud distribution and thickness from all-sky images through star identification and photometry. Specifically, a high-precision star coordinate to the pixel position imaging model calibration method based on the star recognition for fisheye lenses is investigated, resulting in an all-sky rms error of less than 0.87 pixels. Based on the comprehensive reference star catalog, an optimized star extraction method based on SExtractor is developed to handle the difficulty of image source detection in all-sky cloud images. The optical thickness and distribution of cloud layers is calculated through star matching and extinction measurements. Finally, contingent upon the capability of camera and catalog star density, seven cloud layer reconstruction methods are proposed based on meshing and machine learning techniques, achieving a reconstruction accuracy of up to 1.°8. The processing results from real observed images indicate that the proposed method offers a straightforward calibration process and delivers excellent cloud cover extraction and reconstruction outcomes, thereby providing practical value in telescope dynamic scheduling, site characterization and the development of observation strategies.

show abstract

Day and Night Clouds Detection Using a Thermal-Infrared All-Sky-View Camera

Cited by 16 publications

References 36 publications

Twenty-four-hour cloud cover calculation using a ground-based imager with machine learning

Twenty-four-hour cloud cover calculation using a ground-based imager with machine learning

Observations of Nighttime Clouds Over Chiba, Japan, Using Digital Cameras and Satellite Images

Cloud Identification and Reconstruction from All-sky Camera Images Based on Star Photometry Estimation

Contact Info

Product

Resources

About