Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer Learning
Subsidence, especially in populated areas, is becoming a threat to human life and property. Monitoring and analyzing the effects of subsidence over large areas using in situ measurements is difficult and depends on the size of the subsidence area and its location. It is also time-consuming and costl...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/20/7931 |
| _version_ | 1797469932554813440 |
|---|---|
| author | Anna Franczyk Justyna Bała Maciej Dwornik |
| author_facet | Anna Franczyk Justyna Bała Maciej Dwornik |
| author_sort | Anna Franczyk |
| collection | DOAJ |
| description | Subsidence, especially in populated areas, is becoming a threat to human life and property. Monitoring and analyzing the effects of subsidence over large areas using in situ measurements is difficult and depends on the size of the subsidence area and its location. It is also time-consuming and costly. A far better solution that has been used in recent years is Differential Interferometry Synthetic Aperture Radar (DInSAR) monitoring. It allows the monitoring of land deformations in large areas with high accuracy and very good spatial and temporal resolution. However, the analysis of SAR images is time-consuming and involves an expert who can easily overlook certain details. Therefore, it is essential, especially in the case of early warning systems, to prepare tools capable of identifying and monitoring subsidence in interferograms. This article presents a study on automated detection and monitoring of subsidence troughs using deep-transfer learning. The area studied is the Upper Silesian Coal Basin (southern Poland). Marked by intensive coal mining, it is particularly prone to subsidence of various types. Additionally, the results of trough detection obtained with the use of convolutional neural networks were compared with the results obtained with the Hough transform and the circlet transform. |
| first_indexed | 2024-03-09T19:30:55Z |
| format | Article |
| id | doaj.art-a3dac49436204bec9861139c944d225d |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-09T19:30:55Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-a3dac49436204bec9861139c944d225d2023-11-24T02:29:04ZengMDPI AGSensors1424-82202022-10-012220793110.3390/s22207931Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer LearningAnna Franczyk0Justyna Bała1Maciej Dwornik2Department of Geoinformatics and Applied Computer Science, AGH University of Science and Technology, al. Mickiewicza 30, 30-059 Kraków, PolandDepartment of Geoinformatics and Applied Computer Science, AGH University of Science and Technology, al. Mickiewicza 30, 30-059 Kraków, PolandDepartment of Geoinformatics and Applied Computer Science, AGH University of Science and Technology, al. Mickiewicza 30, 30-059 Kraków, PolandSubsidence, especially in populated areas, is becoming a threat to human life and property. Monitoring and analyzing the effects of subsidence over large areas using in situ measurements is difficult and depends on the size of the subsidence area and its location. It is also time-consuming and costly. A far better solution that has been used in recent years is Differential Interferometry Synthetic Aperture Radar (DInSAR) monitoring. It allows the monitoring of land deformations in large areas with high accuracy and very good spatial and temporal resolution. However, the analysis of SAR images is time-consuming and involves an expert who can easily overlook certain details. Therefore, it is essential, especially in the case of early warning systems, to prepare tools capable of identifying and monitoring subsidence in interferograms. This article presents a study on automated detection and monitoring of subsidence troughs using deep-transfer learning. The area studied is the Upper Silesian Coal Basin (southern Poland). Marked by intensive coal mining, it is particularly prone to subsidence of various types. Additionally, the results of trough detection obtained with the use of convolutional neural networks were compared with the results obtained with the Hough transform and the circlet transform.https://www.mdpi.com/1424-8220/22/20/7931subsidence detectionneural networkimage analysis |
| spellingShingle | Anna Franczyk Justyna Bała Maciej Dwornik Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer Learning Sensors subsidence detection neural network image analysis |
| title | Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer Learning |
| title_full | Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer Learning |
| title_fullStr | Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer Learning |
| title_full_unstemmed | Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer Learning |
| title_short | Monitoring Subsidence Area with the Use of Satellite Radar Images and Deep Transfer Learning |
| title_sort | monitoring subsidence area with the use of satellite radar images and deep transfer learning |
| topic | subsidence detection neural network image analysis |
| url | https://www.mdpi.com/1424-8220/22/20/7931 |
| work_keys_str_mv | AT annafranczyk monitoringsubsidenceareawiththeuseofsatelliteradarimagesanddeeptransferlearning AT justynabała monitoringsubsidenceareawiththeuseofsatelliteradarimagesanddeeptransferlearning AT maciejdwornik monitoringsubsidenceareawiththeuseofsatelliteradarimagesanddeeptransferlearning |