Abstract: Based on the observation data of National Basic Weather Stations in Hunan from 2011 to 2020, the temporal and spatial distribution characteristics of low visibility weather in Hunan are studied. The ECMWF model visibility forecast is corrected by using probability density function matching method based on the correction threshold that is obtained by combining the relationship between low visibility and surface meteorological elements. The analysis shows that: (1) The area with low-visibility was mainly located in the central and northern part of Hunan, Tongdao County and Hengyang City. The whole low-visibility area had a obvious tendency to move southwardly. Low visibility occurred most frequently in winter, followed by spring and autumn, and was most widely distributed in winter, concentrated in central and eastern Hunan. (2) The diurnal variation of low visibility was single peak, mainly concentrated from 20:00 to 09:00. (3) The low visibility less than or equal to 1km mainly appeared in the meteorological environment where wind speed was less than 2 m/s, the relative humidity was basically above 90%, the surface temperature was lower than 20°C and the 24-hour pressure change was lower than 2hPa. (4) The visibilty forecast in most low-visibility areas could be corrected by using correction methods that combined probability density function with meteorological factors, which was better than the probability matching correction scheme. After correction, TS scores were significantly improved.