Fog generally forms at dawn and dusk, which exerts serious impacts on public traffic and human health. Terrain strongly affects fog formation, which provides a useful clue for fog detection from satellite observation. With the aid of the advanced Himawari-8 imager data (H8/AHI), this study develops a deep learning algorithm for fog detection at dawn and dusk under terrain-restriction and enhanced channel domain attention mechanism (DDF-Net). The DDF-Net is based on the traditional U-Net model, with the digital elevation model (DEM) data acting as the auxiliary information to separate fog from the low stratus. Furthermore, the squeeze-and-excitation networks (SE-Net) is integrated to optimize the information extraction for eliminating the influence of solar zenith angles (SZA) on the spectral characteristics over a large region. Results show acceptable accuracy of the DDF-Net. The overall probability of detection (POD) is 84.0% at dawn and 83.7% at dusk. In addition, the terrain-restriction strategy improves the results at the edges of foggy regions and reduces the false alarm rate (FAR) for low stratus. The accuracy is expected to be improved when training at a season or month scale, rather than at a longer temporal scale. Results of our study help to improve the accuracy of fog detection, which could further support the relevant traffic planning or healthy travel.

Satellite remote sensing provides a potential technology for detecting fog at dawn and dusk on a large scale. However, the spectral characteristics of fog at dawn and dusk are similar to those of the ground surface, which makes satellite-based fog detection difficult. With the aid of time-series datasets from the Himawari-8 (H8)/AHI, this study proposed a novel algorithm of the self-adaptive threshold of visual background extractor (ST-ViBe) model for satellite fog detection at dawn and dusk. Methodologically, the background model was first built using the difference between MIR and TIR (BTD) and the local binary similarity patterns (LBSP) operator. Second, BTD and scale invariant local ternary pattern (SILTP) texture features were coupled to form scene factors, and the detection threshold of each pixel was determined adaptively to eliminate the influence of the solar zenith angles. The background model was updated rapidly by accelerating the updating rate and increasing the updating quantity. Finally, the residual clouds were removed with the traditional cloud removal method to achieve accurate detection of fog at dawn and dusk over a large area. The validation results demonstrated that the ST-ViBe algorithm could detect fog at dawn and dusk precisely, and on a large scale. The probability of detection, false alarm ratio, and critical success index were 72.5%, 18.5%, 62.4% at dawn (8:00) and 70.6%, 33.6%, 52.3% at dusk (17:00), respectively. Meanwhile, the algorithm mitigated the limitations of the traditional algorithms, such as illumination mutation, missing detection, and residual shadow. The results of this study could guide satellite fog detection at dawn and dusk and improve the detection of similar targets.