Filters

2021 - 20222

More filters

20222
Reset filters

Settings

Author: Banerjee, Abhirup × Author: Kurths, Jürgen × Author: Marwan, Norbert × End date: 2023 × Start date: 2020 × Type: Text ×

Search Results

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Predicting the data structure prior to extreme events from passive observables using echo state network
    (Lausanne : Frontiers Media, 2022) Banerjee, Abhirup; Mishra, Arindam; Dana, Syamal K.; Hens, Chittaranjan; Kapitaniak, Tomasz; Kurths, Jürgen; Marwan, Norbert
    Extreme events are defined as events that largely deviate from the nominal state of the system as observed in a time series. Due to the rarity and uncertainty of their occurrence, predicting extreme events has been challenging. In real life, some variables (passive variables) often encode significant information about the occurrence of extreme events manifested in another variable (active variable). For example, observables such as temperature, pressure, etc., act as passive variables in case of extreme precipitation events. These passive variables do not show any large excursion from the nominal condition yet carry the fingerprint of the extreme events. In this study, we propose a reservoir computation-based framework that can predict the preceding structure or pattern in the time evolution of the active variable that leads to an extreme event using information from the passive variable. An appropriate threshold height of events is a prerequisite for detecting extreme events and improving the skill of their prediction. We demonstrate that the magnitude of extreme events and the appearance of a coherent pattern before the arrival of the extreme event in a time series affect the prediction skill. Quantitatively, we confirm this using a metric describing the mean phase difference between the input time signals, which decreases when the magnitude of the extreme event is relatively higher, thereby increasing the predictability skill.
  • Thumbnail Image
    Item
    Recurrence analysis of extreme event-like data
    (Katlenburg-Lindau : European Geophysical Society, 2021) Banerjee, Abhirup; Goswami, Bedartha; Hirata, Yoshito; Eroglu, Deniz; Merz, Bruno; Kurths, Jürgen; Marwan, Norbert
    The identification of recurrences at various timescales in extreme event-like time series is challenging because of the rare occurrence of events which are separated by large temporal gaps. Most of the existing time series analysis techniques cannot be used to analyze an extreme event-like time series in its unaltered form. The study of the system dynamics by reconstruction of the phase space using the standard delay embedding method is not directly applicable to event-like time series as it assumes a Euclidean notion of distance between states in the phase space. The edit distance method is a novel approach that uses the point-process nature of events. We propose a modification of edit distance to analyze the dynamics of extreme event-like time series by incorporating a nonlinear function which takes into account the sparse distribution of extreme events and utilizes the physical significance of their temporal pattern. We apply the modified edit distance method to event-like data generated from point process as well as flood event series constructed from discharge data of the Mississippi River in the USA and compute their recurrence plots. From the recurrence analysis, we are able to quantify the deterministic properties of extreme event-like data. We also show that there is a significant serial dependency in the flood time series by using the random shuffle surrogate method.