Scholarlink

<p>There are a lot of studies on characterizing the correlation of fluctuation series across different climate variables and quantifying their strength. However, inconsistencies arise when different explicit or implicit assumptions on fluctuation processes are made. To solve this problem, a framework integrating with assumption test on fluctuation processes by means of modified Detrended Fluctuation Analysis (mDFA) is proposed to fully characterize their correlation structure. Both artificially generated series with known deterministic and stochastic components and measured real-world observation series with unknown deterministic and stochastic components are analyzed with this proposed framework to show its efficacy in characterizing the correlation in fluctuations, providing a more accurate fluctuation process representation of underlying series. The importance to build this framework is further highlighted by comparing uncertainties of trend detection from the ground truth fluctuation processes with those from mis-specified ones. All findings given in this study will deepen our understandings on stochastic nature of the measured series in real-world and its potential impacts.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An integrating mDFA-based framework characterizing the correlation in climate fluctuations: with assumed fluctuation processes testing and without trend removal preprocessing

  • 期刊论文

  • 发表日期：

    2026年4月10日

查看全文
  • Shuaifu Lu,
  • Zuntao Fu

摘要

There are a lot of studies on characterizing the correlation of fluctuation series across different climate variables and quantifying their strength. However, inconsistencies arise when different explicit or implicit assumptions on fluctuation processes are made. To solve this problem, a framework integrating with assumption test on fluctuation processes by means of modified Detrended Fluctuation Analysis (mDFA) is proposed to fully characterize their correlation structure. Both artificially generated series with known deterministic and stochastic components and measured real-world observation series with unknown deterministic and stochastic components are analyzed with this proposed framework to show its efficacy in characterizing the correlation in fluctuations, providing a more accurate fluctuation process representation of underlying series. The importance to build this framework is further highlighted by comparing uncertainties of trend detection from the ground truth fluctuation processes with those from mis-specified ones. All findings given in this study will deepen our understandings on stochastic nature of the measured series in real-world and its potential impacts.