To improve the forecast of fog and visibility, prediction experiments are conducted to optimize the two operational diagnostic schemes for forecast of visibility, namely Stoelinga and Warner (SW) scheme and Forecast Systems Laboratory (FSL) scheme. SW scheme takes into account the impact of number concentration of liquid water particles on visibility based on Gultepe scheme. In the improvement of FSL scheme, the decaying averaging method is employed to correct the temperature and dew point and then the relative humidity is recomputed. Base on the simulating results from HUC (hourly update cycle) model of Shandong Institute of Meteorological Sciences, 10 fog processes are selected from 2015 to 2016, among which the forecast of the fog process from 13 to 14 November 2015 is analyzed in detail, and the forecast skills of fog and visibility between the original and improved schemes are compared. The results are as follows. 1) When the rain water content takes up a large proportion in the total liquid water content in the model forecast, the forecast skill of fog and visibility by the improved SW scheme is better than that by the original SW scheme, while when little liquid water content can be forecast by the model, the forecast skill of fog and visibility between the original and improved SW schemes are equivalent. 2) When the mean absolute error (MAE) of the relative humidity, which is recomputed with the corrected temperature and dew point, decreases notably during the forecast period, the forecast skill of fog and visibility by the improved FSL scheme is greatly enhanced. The improved SW and FSL schemes are integrated into one scheme named Combined Visibility (CVIS) scheme, and the results of experiments show the general forecast skill of CVIS scheme is better than that by the two improved schemes.