Page 58 - 水利学报2025年第56卷第4期

.
［ 28］ WU Z， HUANG N E. Ensemble empirical mode decomposition： a noise-assisted data analysis method［J］
Advances in Adaptive Data Analysis，2009，1（1）：1-41.
［ 29］ 郭锐，刘春于，王美洁，等 . 基于平均概率和停止准则的多元 LDPC 码加权符号翻转译码算法［J］ 通信学报，
.
2016，37（2）：43-52.
［ 30］ YANG Q，MA Z，FAN X，et al. Decadal modulation of precipitation patterns over eastern China by sea surface tem ⁃
.
perature anomalies［J］ Journal of Climate，2017，30（17）：7017-7033.
［ 31］ HU S，SPRINTALL J，GUAN C，et al. Deep-reaching acceleration of global mean ocean circulation over the past
two decades［J］ Science Advances，2020，6（6）：eaax7727.
.
.
［ 32］ 丁仁海，丁鑫 . 九华山与周边区域的降水分布差异分析［J］ 气象，2014，40（4）：458-465.
［ 33］ ZHANG W X， ZHOU T J， WU P L. Anthropogenic amplification of precipitation variability over the past century
［J］ Science，2024，385：427-432.
.
.
［ 34］ 李淼，夏军，陈社明，等. 北京地区近300年降水变化的小波分析［J］ 自然资源学报，2011，26（6）：1001-1011.
［ 35］ 龙 笛 ， 杨 文 婷 ， 孙 章 丽 ， 等 . 海 河 平 原 地 下 水 储 量 变 化 的 重 力 卫 星 反 演 和 流 域 水 量 平 衡［J］ 水 利 学 报 ，
.
2023，54（3）：255-267.
.
［ 36］ 陆垂裕，赵勇，刘蓉，等 . 华北平原深层地下水可更新性与水位管控探讨［J］ 中国水利，2024（7）：22-27.
［ 37］ 王庆明，赵勇，王浩，等 . 海河流域地表水资源衰减归因与规律［J］ 中国科学 （地球科学），2024，54 （5）：
.
1573-1587.
［ 38］ 赵勇，常奂宇，桑学锋，等 . 京津冀水资源-粮食-能源-生态协同调控研究Ⅱ：应用［J］ 水利学报， 2022，
.
53（10）：1251-1261.
［ 39］ 赵勇，李海红，刘寒青，等 . 增长的规律：中国用水极值预测［J］ 水利学报，2021，52（2）：129-141.
.

Analysis of 500-year precipitation periodicity and prediction in Haihe River Basin

TAN Weili，ZHAO Yong，WANG Qingming，LIU Rong

（State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin，China Institute of

Water Resources and Hydropower Research，Beijing 100038，China）

Abstract：Periodicity is a fundamental characteristic of precipitation sequences. There has been a long-standing
debate on whether the periodic analysis based on data from station sequences over the past 60 years can represent the

true regional precipitation cycles，making future predictions less convincing. Therefore，it is particularly necessary to

reconstruct precipitation sequences over longer periods and conduct periodic analyses. This study，based on the Long

Short-Term Memory （Long Short-Term Memory，LSTM）algorithm，combined with historical records of drought and

flood levels，as well as station monitoring data，reconstructed a 546-year precipitation sequence in the Haihe River

Basin and analyzed the periodic variations and future trends of precipitation. The main results are as follows. （1）A
drought-flood-precipitation model based on the LSTM algorithm was established，reconstructing the precipitation

sequence in the Haihe River Basin from 1477 to 2022. （2）The 546-year precipitation sequence reveals three signifi⁃

cant cycles in the Haihe River Basin：a 33-year cycle，a 66-year cycle，and a 17-year cycle. Currently，the domi⁃

nant cycle is the 33-year，but the amplitude signal of the 66-year cycle is strengthening. The PDO index shows a 64-

year periodic variation，providing a basis for the intense amplitude of the 66-year precipitation cycle. （3）At present，

precipitation in the Haihe River Basin is in a period of above-average rainfall. According to the 33-year cycle，it is

predicted that there is a 75% probability that the above-average rainfall period in the Haihe River Basin will continue

for 4 to 8 years since 2022，followed by a below-average rainfall period，which is expected to last approximately 15 to

18 years with an 86% probability. Adequate preparedness for water security in the Haihe River Basin is essential to
meet the severe challenges posed by drought cycles.

Keywords：Haihe River Basin；LSTM；precipitation reconstruction；precipitation periodicity；prediction

（责任编辑：耿庆斋）
— 476 —