Abstract: For the sake of identifying the multi scale periodic characteristics of annual and seasonal precipitations and forecasting the precipitation trend in coastal tidal flat region, two kinds of the most popularly used continuous wavelet transforms, Morlet and Mexican Hat wavelet transforms, were conducted based on the long term precipitation series covering 59 years from 1953 to 2011 in coastal Dongtai county. By comparing the two wavelet analysis results,it was proved that Morlet wavelet transform performed better in revealing the local features of coastal precipitation periodicity and increasing the periodic significance level.The results showed that the dominant periodic oscillations involved in coastal annual and seasonal precipitation appeared at inter decadal time scales.The annual and summer precipitation had the similar first,second and third dominant periods of 24,16 and 9 years respectively,and the spring,autumn and winter precipitation had the 16,23 and 17 years dominant periods respectively.The significant precipitation periodicities at interdecadal time scales in the studied area matched the corresponding precipitation periodicity in the whole Jianghuai Area,while the poor significance in precipitation periodicities at inter annual time scales in the studied area didn't match because of the special geography location in coastal region.Further precipitation trend prediction,based on the revealed dominant periodicities in coastal precipitation,showed that the coastal annual and summer precipitation in subsequent 8-12 years, the spring and winter precipitation in subsequent 8 years and the autumn precipitation in subsequent 11-12 years might be in relatively low periods.The present study can be helpful for extending the research field of application mechanism in wavelet analysis,and can provide theoretical basis for the research of climate change and agriculture development in coastal region under the background of coastal reclamation and global climate change.

Key words: Coastal tidal flat region, Morlet wavelet, Mexican Hat wavelet, Wavelet power spectrum, Trend prediction