Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithm

This study addresses the growing challenges of climate extremes and their impact on flood-drought shifts in Xinjiang, China, a region highly sensitive to climate variations. While existing classification models such as logistic regression (LR), support vector machines (SVMs), and geographically weig...

Full description

Saved in:
Bibliographic Details
Main Authors: Sulei Naibi, Anming Bao, Ye Yuan, Jiayu Bao, Rafiq Hamdi, Tao Yu, Xiaoran Huang, Ting Wang, Tao Li, Jingyu Jin, Gang Long, Piet Termonia
Format: Article
Language:English
Published: Elsevier 2025-05-01
Series:Ecological Informatics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S1574954125000391
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849716834403614720
author Sulei Naibi
Anming Bao
Ye Yuan
Jiayu Bao
Rafiq Hamdi
Tao Yu
Xiaoran Huang
Ting Wang
Tao Li
Jingyu Jin
Gang Long
Piet Termonia
author_facet Sulei Naibi
Anming Bao
Ye Yuan
Jiayu Bao
Rafiq Hamdi
Tao Yu
Xiaoran Huang
Ting Wang
Tao Li
Jingyu Jin
Gang Long
Piet Termonia
author_sort Sulei Naibi
collection DOAJ
description This study addresses the growing challenges of climate extremes and their impact on flood-drought shifts in Xinjiang, China, a region highly sensitive to climate variations. While existing classification models such as logistic regression (LR), support vector machines (SVMs), and geographically weighted logistic regression (GWLR) have been applied to spatial data, they exhibit limitations in handling spatial nonstationarity and balancing accuracy with interpretability. To fill this gap, we propose a novel least squares SVM (LSSVM)-based spatially varying coefficient logistic regression (LSSVM-SVCLR) model, which combines the flexibility of LSSVM with the interpretability of logistic regression and the spatial adaptability of spatially varying coefficient models. Through simulations under varying data sizes and complexity, the model achieved high accuracy, with area under the curve (AUC) values approaching 1 in simpler cases and around 0.8 in more complex scenarios. A case study analyzing the relationship between climate extremes and flood-drought shifts in Xinjiang demonstrated the model's applicability, achieving training and testing accuracies of 0.994 and 0.831, respectively, outperforming state-of-the-art machine learning models. Furthermore, the model revealed specific spatial effects of climate extremes on flood-drought shifts, providing probabilistic predictions across the study area. The findings highlight the potential of the proposed model to improve predictions of extreme climate-related events, offering valuable insights for disaster management and climate risk evaluation. This study provides a robust framework for analyzing the complexities of spatial nonstationarity in climate risk analysis.
format Article
id doaj-art-8e8fa28ae833409bbff527e7728fa7ac
institution DOAJ
issn 1574-9541
language English
publishDate 2025-05-01
publisher Elsevier
record_format Article
series Ecological Informatics
spelling doaj-art-8e8fa28ae833409bbff527e7728fa7ac2025-08-20T03:12:51ZengElsevierEcological Informatics1574-95412025-05-018610303010.1016/j.ecoinf.2025.103030Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithmSulei Naibi0Anming Bao1Ye Yuan2Jiayu Bao3Rafiq Hamdi4Tao Yu5Xiaoran Huang6Ting Wang7Tao Li8Jingyu Jin9Gang Long10Piet Termonia11State Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China; University of Chinese Academy of Sciences, Beijing 100049, China; Department of Geography, Ghent University, Ghent 9000, BelgiumState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China; China-Pakistan Joint Research Center on Earth Sciences, Chinese Academy of Sciences and Higher Education Commission, Islamabad 45320, Pakistan; China Key Laboratory of GIS & RS Application Xinjiang Uygur Autonomous Region, Urumqi 830011, China; Corresponding author at: State Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China.State Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, ChinaFaculty of Geographical Science, Beijing Normal University, Beijing 100875, ChinaState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China; Royal Meteorological Institute, Brussels, Belgium; Department of Physics and Astronomy, Ghent University, Ghent, BelgiumState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, ChinaState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China; University of Chinese Academy of Sciences, Beijing 100049, China; Department of Geography, Ghent University, Ghent 9000, BelgiumState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, ChinaState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China; University of Chinese Academy of Sciences, Beijing 100049, ChinaState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China; University of Chinese Academy of Sciences, Beijing 100049, ChinaState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China; University of Chinese Academy of Sciences, Beijing 100049, ChinaRoyal Meteorological Institute, Brussels, Belgium; Department of Physics and Astronomy, Ghent University, Ghent, BelgiumThis study addresses the growing challenges of climate extremes and their impact on flood-drought shifts in Xinjiang, China, a region highly sensitive to climate variations. While existing classification models such as logistic regression (LR), support vector machines (SVMs), and geographically weighted logistic regression (GWLR) have been applied to spatial data, they exhibit limitations in handling spatial nonstationarity and balancing accuracy with interpretability. To fill this gap, we propose a novel least squares SVM (LSSVM)-based spatially varying coefficient logistic regression (LSSVM-SVCLR) model, which combines the flexibility of LSSVM with the interpretability of logistic regression and the spatial adaptability of spatially varying coefficient models. Through simulations under varying data sizes and complexity, the model achieved high accuracy, with area under the curve (AUC) values approaching 1 in simpler cases and around 0.8 in more complex scenarios. A case study analyzing the relationship between climate extremes and flood-drought shifts in Xinjiang demonstrated the model's applicability, achieving training and testing accuracies of 0.994 and 0.831, respectively, outperforming state-of-the-art machine learning models. Furthermore, the model revealed specific spatial effects of climate extremes on flood-drought shifts, providing probabilistic predictions across the study area. The findings highlight the potential of the proposed model to improve predictions of extreme climate-related events, offering valuable insights for disaster management and climate risk evaluation. This study provides a robust framework for analyzing the complexities of spatial nonstationarity in climate risk analysis.http://www.sciencedirect.com/science/article/pii/S1574954125000391Climate extremesClassificationSpatial nonstationarityFlood-drought index
spellingShingle Sulei Naibi
Anming Bao
Ye Yuan
Jiayu Bao
Rafiq Hamdi
Tao Yu
Xiaoran Huang
Ting Wang
Tao Li
Jingyu Jin
Gang Long
Piet Termonia
Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithm
Ecological Informatics
Climate extremes
Classification
Spatial nonstationarity
Flood-drought index
title Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithm
title_full Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithm
title_fullStr Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithm
title_full_unstemmed Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithm
title_short Flood-drought shifts monitoring on arid Xinjiang, China using a novel machine learning based algorithm
title_sort flood drought shifts monitoring on arid xinjiang china using a novel machine learning based algorithm
topic Climate extremes
Classification
Spatial nonstationarity
Flood-drought index
url http://www.sciencedirect.com/science/article/pii/S1574954125000391
work_keys_str_mv AT suleinaibi flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT anmingbao flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT yeyuan flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT jiayubao flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT rafiqhamdi flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT taoyu flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT xiaoranhuang flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT tingwang flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT taoli flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT jingyujin flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT ganglong flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm
AT piettermonia flooddroughtshiftsmonitoringonaridxinjiangchinausinganovelmachinelearningbasedalgorithm