Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithms

Study region: Qarhan Salt Lake, the largest salt lake in China, located in the Qaidam Basin. Study focus: Sustainable management of brine resources in Qarhan Salt Lake is crucial due to the impacts of sustained brine pumping, which has altered brine level and salinity distributions. This study devel...

Full description

Saved in:
Bibliographic Details
Main Authors: Dongmei Yu, Zitao Wang, Chao Yue, Jianping Wang
Format: Article
Language:English
Published: Elsevier 2025-04-01
Series:Journal of Hydrology: Regional Studies
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2214581825000199
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850034697583722496
author Dongmei Yu
Zitao Wang
Chao Yue
Jianping Wang
author_facet Dongmei Yu
Zitao Wang
Chao Yue
Jianping Wang
author_sort Dongmei Yu
collection DOAJ
description Study region: Qarhan Salt Lake, the largest salt lake in China, located in the Qaidam Basin. Study focus: Sustainable management of brine resources in Qarhan Salt Lake is crucial due to the impacts of sustained brine pumping, which has altered brine level and salinity distributions. This study developed an automated machine learning (AutoML) approach to model brine levels and salinity, providing a tool for informed resource management decisions. The Geodetector was employed to quantify the influence of various factors on these parameters. New hydrological insights for the region: An integrated approach using AutoML and GIS significantly improved prediction accuracy for both brine levels and salinity. For brine level prediction, the LightGBM (LGBM) model performed best, achieving an R2 of 0.880 (training) and 0.869 (testing). For salinity, Random Forest (RF) was optimal, with an R2 of 0.895 (training) and 0.881 (testing). Geodetector analysis revealed that distance to pumps (q = 0.544), canal density (q = 0.346), lithology (q = 0.324), and distance to lakes (q = 0.260) are key factors influencing brine levels. For salinity, precipitation (q = 0.350) and distance to lakes (q = 0.097) were found to be the most influential. This study demonstrates AutoML's effectiveness in modeling brine dynamics and offers insights into factors influencing changes, aiding brine extraction optimization and sustainable resource management in fragile salt lake ecosystems.
format Article
id doaj-art-8edf64776c04455f9485f540fcfb2cc1
institution DOAJ
issn 2214-5818
language English
publishDate 2025-04-01
publisher Elsevier
record_format Article
series Journal of Hydrology: Regional Studies
spelling doaj-art-8edf64776c04455f9485f540fcfb2cc12025-08-20T02:57:44ZengElsevierJournal of Hydrology: Regional Studies2214-58182025-04-015810219510.1016/j.ejrh.2025.102195Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithmsDongmei Yu0Zitao Wang1Chao Yue2Jianping Wang3Key Laboratory of Comprehensive and Highly Efficient Utilization of Salt Lake Resources, Qinghai Institute of Salt Lakes, Chinese Academy of Sciences, Xining 810008, China; University of Chinese Academy of Sciences, Beijing 100049, ChinaSchool of Earth and Environment, Anhui University of Science and Technology. Huainan 232001, China; Corresponding author.Key Laboratory of Comprehensive and Highly Efficient Utilization of Salt Lake Resources, Qinghai Institute of Salt Lakes, Chinese Academy of Sciences, Xining 810008, China; University of Chinese Academy of Sciences, Beijing 100049, ChinaKey Laboratory of Comprehensive and Highly Efficient Utilization of Salt Lake Resources, Qinghai Institute of Salt Lakes, Chinese Academy of Sciences, Xining 810008, ChinaStudy region: Qarhan Salt Lake, the largest salt lake in China, located in the Qaidam Basin. Study focus: Sustainable management of brine resources in Qarhan Salt Lake is crucial due to the impacts of sustained brine pumping, which has altered brine level and salinity distributions. This study developed an automated machine learning (AutoML) approach to model brine levels and salinity, providing a tool for informed resource management decisions. The Geodetector was employed to quantify the influence of various factors on these parameters. New hydrological insights for the region: An integrated approach using AutoML and GIS significantly improved prediction accuracy for both brine levels and salinity. For brine level prediction, the LightGBM (LGBM) model performed best, achieving an R2 of 0.880 (training) and 0.869 (testing). For salinity, Random Forest (RF) was optimal, with an R2 of 0.895 (training) and 0.881 (testing). Geodetector analysis revealed that distance to pumps (q = 0.544), canal density (q = 0.346), lithology (q = 0.324), and distance to lakes (q = 0.260) are key factors influencing brine levels. For salinity, precipitation (q = 0.350) and distance to lakes (q = 0.097) were found to be the most influential. This study demonstrates AutoML's effectiveness in modeling brine dynamics and offers insights into factors influencing changes, aiding brine extraction optimization and sustainable resource management in fragile salt lake ecosystems.http://www.sciencedirect.com/science/article/pii/S2214581825000199Spatial modelingBrine levelSalinityQarhan Salt LakeAutomated machine learningGeodetector
spellingShingle Dongmei Yu
Zitao Wang
Chao Yue
Jianping Wang
Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithms
Journal of Hydrology: Regional Studies
Spatial modeling
Brine level
Salinity
Qarhan Salt Lake
Automated machine learning
Geodetector
title Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithms
title_full Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithms
title_fullStr Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithms
title_full_unstemmed Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithms
title_short Spatial modeling of brine level and salinity in the Qarhan Salt Lake using GIS and automated machine learning algorithms
title_sort spatial modeling of brine level and salinity in the qarhan salt lake using gis and automated machine learning algorithms
topic Spatial modeling
Brine level
Salinity
Qarhan Salt Lake
Automated machine learning
Geodetector
url http://www.sciencedirect.com/science/article/pii/S2214581825000199
work_keys_str_mv AT dongmeiyu spatialmodelingofbrinelevelandsalinityintheqarhansaltlakeusinggisandautomatedmachinelearningalgorithms
AT zitaowang spatialmodelingofbrinelevelandsalinityintheqarhansaltlakeusinggisandautomatedmachinelearningalgorithms
AT chaoyue spatialmodelingofbrinelevelandsalinityintheqarhansaltlakeusinggisandautomatedmachinelearningalgorithms
AT jianpingwang spatialmodelingofbrinelevelandsalinityintheqarhansaltlakeusinggisandautomatedmachinelearningalgorithms