Hyeongmok Lee , Go-Eun Kim , Woo-Jin Shin , Yuyoung Lee , Sanghee Park , Kwang-Sik Lee , Jina Jeong , Seung-Ik Park , Sungwook Choung
{"title":"Integrating geostatistical methods and deep learning for enhanced 87Sr/86Sr isoscape Estimation: A case study in South Korea","authors":"Hyeongmok Lee , Go-Eun Kim , Woo-Jin Shin , Yuyoung Lee , Sanghee Park , Kwang-Sik Lee , Jina Jeong , Seung-Ik Park , Sungwook Choung","doi":"10.1016/j.jag.2025.104697","DOIUrl":null,"url":null,"abstract":"<div><div>The <sup>87</sup>Sr/<sup>86</sup>Sr isotopic ratio has emerged as a valuable geochemical tracer in fields such as environmental forensics, archaeology, and provenance research. However, generating accurate and spatially continuous isoscape maps from sparse isotopic measurements remains a major challenge due to limited data availability and spatial heterogeneity. To address this, we propose a hybrid framework for <sup>87</sup>Sr/<sup>86</sup>Sr isoscape mapping that integrates a kriging-based data augmentation method with a deep learning (DL) classifier. The kriging component generates synthetic training samples by interpolating sparse isotopic data while preserving underlying spatial correlations and geological anisotropy. These augmented data, along with spatial geological features (e.g., lithology, tectonic settings) and geochemical compositions, are used as input variables for training a feedforward deep neural network. The approach was applied to 409 soil samples collected across South Korea, and its performance was benchmarked against conventional kriging and convolutional neural networks (CNN). The proposed model achieved significantly higher classification accuracy (91.67%) compared to kriging-based and CNN-based models (76.7% and 86.7%, respectively). Furthermore, the isoscape outputs revealed meaningful isotopic patterns linked to geological and geomorphological controls, such as metamorphic rock distributions, fault density, and surface slope. This framework demonstrates the effectiveness of combining geostatistics with DL to improve predictive accuracy and interpretability in isotopic provenance research and environmental monitoring.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"142 ","pages":"Article 104697"},"PeriodicalIF":8.6000,"publicationDate":"2025-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"International journal of applied earth observation and geoinformation : ITC journal","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1569843225003449","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"REMOTE SENSING","Score":null,"Total":0}
引用次数: 0
Abstract
The 87Sr/86Sr isotopic ratio has emerged as a valuable geochemical tracer in fields such as environmental forensics, archaeology, and provenance research. However, generating accurate and spatially continuous isoscape maps from sparse isotopic measurements remains a major challenge due to limited data availability and spatial heterogeneity. To address this, we propose a hybrid framework for 87Sr/86Sr isoscape mapping that integrates a kriging-based data augmentation method with a deep learning (DL) classifier. The kriging component generates synthetic training samples by interpolating sparse isotopic data while preserving underlying spatial correlations and geological anisotropy. These augmented data, along with spatial geological features (e.g., lithology, tectonic settings) and geochemical compositions, are used as input variables for training a feedforward deep neural network. The approach was applied to 409 soil samples collected across South Korea, and its performance was benchmarked against conventional kriging and convolutional neural networks (CNN). The proposed model achieved significantly higher classification accuracy (91.67%) compared to kriging-based and CNN-based models (76.7% and 86.7%, respectively). Furthermore, the isoscape outputs revealed meaningful isotopic patterns linked to geological and geomorphological controls, such as metamorphic rock distributions, fault density, and surface slope. This framework demonstrates the effectiveness of combining geostatistics with DL to improve predictive accuracy and interpretability in isotopic provenance research and environmental monitoring.
期刊介绍:
The International Journal of Applied Earth Observation and Geoinformation publishes original papers that utilize earth observation data for natural resource and environmental inventory and management. These data primarily originate from remote sensing platforms, including satellites and aircraft, supplemented by surface and subsurface measurements. Addressing natural resources such as forests, agricultural land, soils, and water, as well as environmental concerns like biodiversity, land degradation, and hazards, the journal explores conceptual and data-driven approaches. It covers geoinformation themes like capturing, databasing, visualization, interpretation, data quality, and spatial uncertainty.