Time series analysis of red tide's disaster characteristics in China seas

基于时间序列的海洋赤潮灾害特征分析（英文）

The annual frequency of red tides from 1977 to 2012 and the monthly frequency of red tides from 2001 to 2012 in China seas were used to establish the time series of red tide annual frequency and monthly frequency, respectively. The annual frequency fit well with time segments revealed by piecewise linear regression analysis. The seasonal maximum of monthly frequency was in May （-18.22）, and the stochastic volatility tended to increase gradually with time series, with peak values occurring from May to July. Holt exponential smoothing and Holt-winter exponential smoothing were used to predict red tide annual and monthly frequencies, which revealed that the annual frequency of red tides would rise slowly by one time from 2013 to 2020, and that red tides would mainly occur from May to July in 2013-2016 with a peak value of about 25 times in May.

以1977-2012年中国海赤潮的年发生频率及2001-2012年赤潮的月发生频率数据资料为基础,建立赤潮事件的年发生频率和月发生频率时间序列。赤潮的年发生频率与时间的分段回归拟合效果较好,月频率的季节性最大值在5月(约18.22),随机波动的大小随时间序列逐步增加,波动峰值主要出现在5-7月。利用Holt指数平滑法和Holt-Winter指数平滑法分别对赤潮事件的年发生频率和月发生频率进行预测,结果表明2013-2020年赤潮的年发生频率呈年平均增加1次的缓慢趋势上升,2013-2016年5-7月份为赤潮高发期,峰值出现在5月,基本稳定在25次左右。