湖泊科学2026，Vol.38Issue(5)：1899-1908,10.DOI:10.18307/2026.0515
水库浮游植物自动监测数据解析:利用多周期时序分解方法解析群落结构多尺度动态
Analysis of phytoplankton automatic monitoring data in reservoirs:Deciphering multi-scale dynamics of community structure using multi-period time series decomposition methods
摘要
Abstract
Exploring the temporal variation characteristics of phytoplankton community structure is of great significance for under-standing the evolution of the aquatic ecological environment.Automatic monitoring of phytoplankton communities provides funda-mental data for revealing the multi-scale changes in phytoplankton community structure.However,high-frequency monitoring data often exhibit characteristics of multiple period superposition and frequent occurrence of outliers,which poses challenges to tradition-al time series analysis methods in terms of unstable period decomposition.This study takes Jiangdong Reservoir in Beixi of the Jiu-long River as a case study,and obtains hourly automatic monitoring data on the biomass of four phytoplankton groups(Chlorophy-ta,Cyanobacteria,Bacillariophyta,and Cryptophyta)from 2017 to 2022.Using multiple seasonal-trend decomposition using LO-ESS(MSTL),the biomass of each group was decomposed into four components:trend,annual period,diurnal rhythm,and resid-ual.This method enables the quantitative characterization of the variation characteristics of phytoplankton communities at different time scales(especially diurnal rhythms)based on years of continuous high-frequency monitoring data within the same analytical framework.The results showed that:(1)The biomass of the Chlorophyta group increased rapidly from mid-2019,reached its peak in 2021,and became the dominant group;the overall trend of the other groups was declining.(2)All groups exhibited obvious an-nual periodic characteristics,and their intra-annual fluctuations showed good consistency across different years.Among them,Cryptophyta had the largest fluctuation amplitude,followed by Chlorophyta and Cyanobacteria,while Bacillariophyta had the smal-lest.(3)All groups had stable diurnal rhythms,with their peaks occurring between 12:00-16:00,Chlorophyta showing the lar-gest amplitude,and Cyanobacteria showing the smoothest changes.(4)Seasonal cycles and long-term trends jointly accounted for the main part of the variation in community biomass,while the contribution of diurnal cycles was relatively small.The research re-sults indicate that the changes in Chlorophyta and Bacillariophyta groups are driven by both long-term background and seasonal fac-tors,while the fluctuations in Cyanobacteria and Cryptophyta groups are more dependent on seasonal temperature and light cycles.This study clarifies the dynamic patterns of phytoplankton at annual,seasonal,and diurnal scales,as well as the contribution weights of each scale,which can provide a scientific basis for the early warning of harmful algal blooms and the management of a-quatic ecosystems in subtropical monsoon climate reservoirs.关键词
浮游植物/多周期时间序列分解方法(MSTL)/趋势演替/群落结构/多尺度分析Key words
Phytoplankton/multiple seasonal-trend decomposition using LOESS(MSTL)/trend succession/community structure/multiscale analysis引用本文复制引用
李佳,舒雅琦,梁中耀,陈纪新,陈能汪,刘永..水库浮游植物自动监测数据解析:利用多周期时序分解方法解析群落结构多尺度动态[J].湖泊科学,2026,38(5):1899-1908,10.基金项目
国家自然科学基金项目(52409106)和福建省自然科学基金项目(2024J01047)联合资助. （52409106）
