Abstract: According the automatic observation of ground-based cloud cover in complex meteorological scenarios,this study proposed a dynamic multi-threshold cloud cover identification method driven by solar altitude and visibility.Based on a database of 190 628 ground-based cloud image samples from 24 representative stations across China from 2021 to 2023,a machine learning approach was employed to construct an adaptive multi-threshold generation model by fusing visible light radiation characteristics of ground-based cloud images,real-time meteorological optical range,and solar altitude angle parameters.This model breaks through the adaptability bottleneck of traditional fixed threshold methods in scenarios such as dawn/dusk,thin clouds,and low visibility.The results show that the cloud-sky segmentation threshold increases with the rise of solar altitude angle,with threshold ranges of 0.95-1.15 and 1.30-1.45 for low and high solar altitude angles,respectively.The cloud-sky segmentation threshold decreases as visibility decreases; when visibility is >10 km or <1 km,the solar altitude angle threshold ranges are 1.07-1.45 and 0.95-1.15,respectively,while the threshold segmentation method loses reliability when visibility is <500 m.Validation through interactive scenario threshold verification and cloud type differentiation verification demonstrates that this identification method reduces the mean absolute error of cloud cover identification to 11.3% in complex scenarios with <10° solar altitude and <10 km visibility,representing a 22.0% improvement over the fixed threshold method.The recognition accuracy rates of cloud cover at 0%-10%,10%-30%,30%-70%,and>70% reach 93.8%,89.5%,95.3%,and 93.7% respectively,which are 3.6%,6.8%,2.2%,and 2.5% higher than those of the existing optimal baseline (DeepLab V3+) method.