ABSTRACT
Predicting precipitation δ18O accurately is crucial for understanding water cycles, paleoclimates, and hydrological applications. Yet, forecasting its spatio-temporal distribution remains challenging due to complex climate interactions and extreme events. We developed a method combining spatio-temporal clustering and deep learning neural networks to improve multi-site, multi-year precipitation δ18O predictions. Using a comprehensive dataset from 33 German sites (1978-2012), our model considers precipitation δ18O and its controlling factors, including precipitation and temperature distribution. We applied the K-means++ method for classification and divided data into training and prediction sets. The CNN (Convolutional Neural Network) model extracted spatial features, while the Bi-LSTM (Bi-directional Long Short-Term Memory) model focused on temporal features. Spatio-temporal clustering using K-means++ improved forecast accuracy and reduced errors. This study highlights the potential of deep learning and clustering techniques for forecasting complex spatio-temporal data and offers insights for future research on isotope distributions.
Disclaimer
As a service to authors and researchers we are providing this version of an accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proofs will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to these versions also.CRediT authorship contribution statement
The research project was jointly conceived by YL and SYH. YL conducted the research questions, data collection, collation, and formal data analysis, and the acquisition of research funding was arranged by SYH. YL and SYH developed the research methods, and investigations were conducted by YL and BBP. The project was managed by SYH and supervised by YL and BM. YL completed all tasks related to data, concepts, and visualization of results, as well as the writing of scripts for back trajectory visualization. YL authored the complete original draft of the paper, which was subsequently edited and reviewed by YL, SYH, BM, JL, BBP, QKT, LBY, QG and DW.
Data Availability Statement
Data archived in a repository: The [Precipitation isotopes, temperature, and precipitation in Germany] data used in the study [4 Results] can be [freely accessible] in [Datasets-Group-GNIP-Monthly] by registering an account at [https://nucleus.iaea.org/wiser/index.aspx].
The [CNN, Bi-LSTM, and CNN-Bi-LSTM model codes] used in the study can be [freely accessed] in [Precipitation-isotope-prediction-model] through the repository search after registering an account at [Github].
Software Availability Statement:
The [Matlab 2021a] used to run the [4 Results prediction model code] is available through the [Products section after registering an account at https://ww2.mathworks.cn/?s_tid=gn_logo].
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/02626667.2024.2375403