In this study, a hybrid nighttime fog detection algorithm was developed based on optical and textural characteristics of fog from Himawari-8 / AHI(advanced himawari imager) data. The dual channel difference(DCD) method based on the difference in emissivity between the 3.9 and 11.2 μm was the main evaluation element for fog detection. Additionally, to distinguish between fog and low cloud, the normalized local standard deviation of brightness temperature and difference between fog top brightness temperature and ground surface temperature (sea surface temperature) were also used. The thresholds and weights of the evaluation elements were optimized through a ROC(receiver operating characteristics) analysis using training cases and visibility meter data. The quantitative evaluation of fog detection using ground observation data, the average POD(probability of detection) and FAR(false alarm ratio) for the evaluation cases were 0.77 and 0.63, respectively, and TS(threat score) and Bias were 0.33 and 2.10, respectively. Detection levels differ depending on fog characteristics(intensity and weather condition) and season. We performed sensitivity tests for the optimization and fog detection methods because the detection levels can be affected by the optimized method and fog detection method. The sensitivity analysis showed that the final optimized threshold values by the TS maximum and by the ROC analysis were exactly same. And the Weighted sum method showed a slightly lowered detection level compared to that of the removal method, average differences(Weighted sum method ? Removal method) of evaluation elements, KSS, TS, and Bias are ?0.04, -0.16, and 3.12, respectively. Although the differences are not positive, we need a more test for the Weighted sum method because it is sensitive to the threshold and weight shape using more cases. And the optimization of threshold values based on the season showed different results according to season. So, more works are needed for the improvement of fog detection levels through the sophistication of threshold values and weight values of test elements.