{"title":"利用混合和集合深度学习模型探测地面沉降","authors":"Narges Kariminejad, Aliakbar Mohammadifar, Adel Sepehr, Mohammad Kazemi Garajeh, Mahrooz Rezaei, Gloria Desir, Adolfo Quesada-Román, Hamid Gholami","doi":"10.1007/s12518-024-00572-9","DOIUrl":null,"url":null,"abstract":"<div><p>Land subsidence (LS) is among the most prominent forms of subsurface erosion and geomorphological hazards. This study used two deep learning (DL) models consisting of the hybrid CNN-RNN and ensemble DL (EDL) merged with two dense models. The main variables controlling LS (consisting of environmental, hydrological, hydrogeological, digital elevation model, and soil characteristics), were used as the input for the predictive DL models. Likewise, to establish the degree of performance of each parameter, different control points have been established. We then trained and tested our DL models using the receiver-operating characteristic-area under curve (ROC-AUC) and precision-recall plots. The measures based on the game theory consisting of permutation feature importance measure (PFIM) and SHapley Additive exPlanations (SHAP) were employed to assess the features relative importance and interpretability of the predictive model output. Our findings show that the ensemble CNN-RNN model performed well with the ROC-AUC curve (0.95) of class 1 (land subsidence) for training data for detecting and mapping land subsidence compared to EDL with the ROC curve (0.93) of class 1 (land subsidence) for training datasets. The CNN-RNN also performed well with the precision-recall curve (0.954) of class 1 for testing data for detecting and mapping land subsidence compared to the EDL model with the precision-recall curve (0.95) of class 1. The results of this research revealed that much of the study area is susceptible to land subsidence. The results of the model sensitivity analysis suggested that the groundwater drop rate is the most sensitive for the model. Based on the SHAP values, the groundwater drop rate was identified as the most contributed feature to the model output. The importance of this study is at a broader level, especially in arid and semiarid environments with similar geomorphological, and climatic conditions.</p></div>","PeriodicalId":46286,"journal":{"name":"Applied Geomatics","volume":null,"pages":null},"PeriodicalIF":2.3000,"publicationDate":"2024-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Detection of land subsidence using hybrid and ensemble deep learning models\",\"authors\":\"Narges Kariminejad, Aliakbar Mohammadifar, Adel Sepehr, Mohammad Kazemi Garajeh, Mahrooz Rezaei, Gloria Desir, Adolfo Quesada-Román, Hamid Gholami\",\"doi\":\"10.1007/s12518-024-00572-9\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><p>Land subsidence (LS) is among the most prominent forms of subsurface erosion and geomorphological hazards. This study used two deep learning (DL) models consisting of the hybrid CNN-RNN and ensemble DL (EDL) merged with two dense models. The main variables controlling LS (consisting of environmental, hydrological, hydrogeological, digital elevation model, and soil characteristics), were used as the input for the predictive DL models. Likewise, to establish the degree of performance of each parameter, different control points have been established. We then trained and tested our DL models using the receiver-operating characteristic-area under curve (ROC-AUC) and precision-recall plots. The measures based on the game theory consisting of permutation feature importance measure (PFIM) and SHapley Additive exPlanations (SHAP) were employed to assess the features relative importance and interpretability of the predictive model output. Our findings show that the ensemble CNN-RNN model performed well with the ROC-AUC curve (0.95) of class 1 (land subsidence) for training data for detecting and mapping land subsidence compared to EDL with the ROC curve (0.93) of class 1 (land subsidence) for training datasets. The CNN-RNN also performed well with the precision-recall curve (0.954) of class 1 for testing data for detecting and mapping land subsidence compared to the EDL model with the precision-recall curve (0.95) of class 1. The results of this research revealed that much of the study area is susceptible to land subsidence. The results of the model sensitivity analysis suggested that the groundwater drop rate is the most sensitive for the model. Based on the SHAP values, the groundwater drop rate was identified as the most contributed feature to the model output. The importance of this study is at a broader level, especially in arid and semiarid environments with similar geomorphological, and climatic conditions.</p></div>\",\"PeriodicalId\":46286,\"journal\":{\"name\":\"Applied Geomatics\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":2.3000,\"publicationDate\":\"2024-07-08\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Applied Geomatics\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://link.springer.com/article/10.1007/s12518-024-00572-9\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"REMOTE SENSING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Applied Geomatics","FirstCategoryId":"1085","ListUrlMain":"https://link.springer.com/article/10.1007/s12518-024-00572-9","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"REMOTE SENSING","Score":null,"Total":0}
Detection of land subsidence using hybrid and ensemble deep learning models
Land subsidence (LS) is among the most prominent forms of subsurface erosion and geomorphological hazards. This study used two deep learning (DL) models consisting of the hybrid CNN-RNN and ensemble DL (EDL) merged with two dense models. The main variables controlling LS (consisting of environmental, hydrological, hydrogeological, digital elevation model, and soil characteristics), were used as the input for the predictive DL models. Likewise, to establish the degree of performance of each parameter, different control points have been established. We then trained and tested our DL models using the receiver-operating characteristic-area under curve (ROC-AUC) and precision-recall plots. The measures based on the game theory consisting of permutation feature importance measure (PFIM) and SHapley Additive exPlanations (SHAP) were employed to assess the features relative importance and interpretability of the predictive model output. Our findings show that the ensemble CNN-RNN model performed well with the ROC-AUC curve (0.95) of class 1 (land subsidence) for training data for detecting and mapping land subsidence compared to EDL with the ROC curve (0.93) of class 1 (land subsidence) for training datasets. The CNN-RNN also performed well with the precision-recall curve (0.954) of class 1 for testing data for detecting and mapping land subsidence compared to the EDL model with the precision-recall curve (0.95) of class 1. The results of this research revealed that much of the study area is susceptible to land subsidence. The results of the model sensitivity analysis suggested that the groundwater drop rate is the most sensitive for the model. Based on the SHAP values, the groundwater drop rate was identified as the most contributed feature to the model output. The importance of this study is at a broader level, especially in arid and semiarid environments with similar geomorphological, and climatic conditions.
期刊介绍:
Applied Geomatics (AGMJ) is the official journal of SIFET the Italian Society of Photogrammetry and Topography and covers all aspects and information on scientific and technical advances in the geomatics sciences. The Journal publishes innovative contributions in geomatics applications ranging from the integration of instruments, methodologies and technologies and their use in the environmental sciences, engineering and other natural sciences.
The areas of interest include many research fields such as: remote sensing, close range and videometric photogrammetry, image analysis, digital mapping, land and geographic information systems, geographic information science, integrated geodesy, spatial data analysis, heritage recording; network adjustment and numerical processes. Furthermore, Applied Geomatics is open to articles from all areas of deformation measurements and analysis, structural engineering, mechanical engineering and all trends in earth and planetary survey science and space technology. The Journal also contains notices of conferences and international workshops, industry news, and information on new products. It provides a useful forum for professional and academic scientists involved in geomatics science and technology.
Information on Open Research Funding and Support may be found here: https://www.springernature.com/gp/open-research/institutional-agreements