Abstract:

Based on daily precipitation data from the 78 meteorological stations in Heilongjiang province from 1971 to 2016, this paper analyzed the historical and future trends of precipitation change during the crop growing season (from May to September) in Heilongjiang province by using the methods of Mexican-hat wavelet transform and Hurst exponent.The results show that there are the major quasi-periods of 7 years, 14 years and 7 years in the growing season, May and June, respectively, and the first quasi-periods of 2 years, 3 years and 7 years as well as the second quasi-periods of 6 years, 11 years and 21 years in July, August and September, respectively, for precipitation in Heilongjiang province during 1971-2016.Meanwhile, the precipitation in each month shows an increasing trend in recent years.The precipitation in growing season shows a fluctuant interannual variation and the change amplitudes are relatively large in July and August.In addition, the precipitation experiences an interdecadal change of increase-decrease-increase pattern.Specifically, the precipitation is above the normal level in the 1980s and 1990s and sharply increases since 2010.The Hurst exponent of each month is above 0.5 at each station, which means that there is an obvious Hurst Phenomenon for precipitation.Precipitation mainly occurs in summer, especially in July.The facts that the precipitation increases and concentrates abnormally in July in recent several years as well as persistently increases in main river basins in July in the future are worthy of attention in agricultural disaster prevention and mitigation.

Key words: Precipitation time series, Mexican-hat wavelet transform, Hurst exponent