Abstract: Based on observed data,including visibility,relative humidity and weather phenomena,from Huzhou national basic meteorological station from 1961 to 2011,the haze days data was reconstructed by using of two daily average methods.The optimal reconstructed result was analyzed.The results showed that the meteorological data of haze days reconstructed by the daily average method had certain reference value.Among which 3 times a day on average as the average daily value to reconstruct the meteorological data sequence was welldistributed and accorded with the changing rules of weather phenomenon.And it had a good correlation and correspondence degree with the records from surface meteorological observed data.The haze days increased obviously,and haze days on different levels increased as well.The severe haze days were increased significantly.The haze days and its different levels were highest in winter and lowest in summer.Haze days in four seasons showed increasing trend,among which the days in autumn increased obviously,followed by the winter and summer was slowest.Monthly variation of haze days and haze days on different levels showed a"V"distribution,and the peak value appeared in December and the bottom was in July.Haze days in each month showed increasing trend,and in October particular.The research results could provide a basis for scientific understanding of haze and some references for ecological construction and atmospheric environmental governance.

Key words: Haze days, Data reconstruction, Haze levels, Huzhou, Haze