Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices

Anas Hesham; Dursun Zafer Seker

doi:10.1007/978-3-031-43218-7_45

Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices

Anas Hesham^*, Dursun Zafer Seker

^*Corresponding author for this work

Department of Geomatics Engineering

Istanbul Technical University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Water bodies like rivers and lakes are essential human, animal, and vegetation resources. Acquiring and analyzing these data are essential for better water resources and environmental management. In recent decades, many methods have been used to extract water bodies from remote sensing images. A critical way was using water indices such as normalized difference water index (NDWI) and modified normalized difference water index (MNDWI). Furthermore, extracting small water bodies (such as creeks) from satellite images is challenging and needs high spatial resolution images. However, there are several difficulties in using water indices in high spatial resolution images because of the limited spectral resolution. Most high spatial resolution satellites lack SWIR, which is necessary to calculate MNDWI. But, WorldView-3 overcomes this problem by having NIR and SWIR bands with high spatial resolution. This study used a Kaggle dataset called DSTL, which includes WorldView-3 images for the same region (but the dataset's provider obscures the location of these images) with 0.31 m, 1.24 m, and 7.5 m spatial resolution at nadir for panchromatic, multi-spectral, and SWIR, respectively. A convolutional neural network (CNN) was proposed to improve water body extraction from satellite images. In this network, three indices (normalized difference vegetation index (NDVI), NDWI, and MNDWI) were input layers to U-Net architecture. To evaluate this approach, the results were compared with the same U-Net architecture using RGB and all WorldView-3 bands as input layers, SegNet, and FCN-8. The Jaccard index was calculated as 90.7%, 90%, 89.3%, 85%, and 80% for U-Net with three indices, RGB, all WorldView-3 bands, SegNet, and FCN-8, respectively. The results show that using indices as input layers to U-Net improves the water body extraction, which is necessary to be accurate in several hydrological and water resources studies.

Original language	English
Title of host publication	Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3
Editors	Attila Çiner, Zeynal Abiddin Ergüler, Mourad Bezzeghoud, Mustafa Ustuner, Mehdi Eshagh, Hesham El-Askary, Arkoprovo Biswas, Luca Gasperini, Klaus-Günter Hinzen, Murat Karakus, Cesare Comina, Ali Karrech, Alina Polonia, Helder I. Chaminé
Publisher	Springer Nature
Pages	193-196
Number of pages	4
ISBN (Print)	9783031432170
DOIs	https://doi.org/10.1007/978-3-031-43218-7_45
Publication status	Published - 2024
Event	1st International conference on Mediterranean Geosciences Union, MedGU 2021 - Istanbul, Turkey Duration: 25 Nov 2021 → 28 Nov 2021

Publication series

Name	Advances in Science, Technology and Innovation
ISSN (Print)	2522-8714
ISSN (Electronic)	2522-8722

Conference

Conference	1st International conference on Mediterranean Geosciences Union, MedGU 2021
Country/Territory	Turkey
City	Istanbul
Period	25/11/21 → 28/11/21

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

Keywords

Convolutional neural network
Deep learning
Remote sensing
Water body extraction
Water index
WorldView-3

Access to Document

10.1007/978-3-031-43218-7_45

Cite this

Hesham, A., & Seker, D. Z. (2024). Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices. In A. Çiner, Z. A. Ergüler, M. Bezzeghoud, M. Ustuner, M. Eshagh, H. El-Askary, A. Biswas, L. Gasperini, K.-G. Hinzen, M. Karakus, C. Comina, A. Karrech, A. Polonia, & H. I. Chaminé (Eds.), Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3 (pp. 193-196). (Advances in Science, Technology and Innovation). Springer Nature. https://doi.org/10.1007/978-3-031-43218-7_45

Hesham, Anas ; Seker, Dursun Zafer. / Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices. Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3. editor / Attila Çiner ; Zeynal Abiddin Ergüler ; Mourad Bezzeghoud ; Mustafa Ustuner ; Mehdi Eshagh ; Hesham El-Askary ; Arkoprovo Biswas ; Luca Gasperini ; Klaus-Günter Hinzen ; Murat Karakus ; Cesare Comina ; Ali Karrech ; Alina Polonia ; Helder I. Chaminé. Springer Nature, 2024. pp. 193-196 (Advances in Science, Technology and Innovation).

@inproceedings{edad7485b0de480e83656d21a88c2fd9,

title = "Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices",

abstract = "Water bodies like rivers and lakes are essential human, animal, and vegetation resources. Acquiring and analyzing these data are essential for better water resources and environmental management. In recent decades, many methods have been used to extract water bodies from remote sensing images. A critical way was using water indices such as normalized difference water index (NDWI) and modified normalized difference water index (MNDWI). Furthermore, extracting small water bodies (such as creeks) from satellite images is challenging and needs high spatial resolution images. However, there are several difficulties in using water indices in high spatial resolution images because of the limited spectral resolution. Most high spatial resolution satellites lack SWIR, which is necessary to calculate MNDWI. But, WorldView-3 overcomes this problem by having NIR and SWIR bands with high spatial resolution. This study used a Kaggle dataset called DSTL, which includes WorldView-3 images for the same region (but the dataset's provider obscures the location of these images) with 0.31 m, 1.24 m, and 7.5 m spatial resolution at nadir for panchromatic, multi-spectral, and SWIR, respectively. A convolutional neural network (CNN) was proposed to improve water body extraction from satellite images. In this network, three indices (normalized difference vegetation index (NDVI), NDWI, and MNDWI) were input layers to U-Net architecture. To evaluate this approach, the results were compared with the same U-Net architecture using RGB and all WorldView-3 bands as input layers, SegNet, and FCN-8. The Jaccard index was calculated as 90.7%, 90%, 89.3%, 85%, and 80% for U-Net with three indices, RGB, all WorldView-3 bands, SegNet, and FCN-8, respectively. The results show that using indices as input layers to U-Net improves the water body extraction, which is necessary to be accurate in several hydrological and water resources studies.",

keywords = "Convolutional neural network, Deep learning, Remote sensing, Water body extraction, Water index, WorldView-3",

author = "Anas Hesham and Seker, {Dursun Zafer}",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; 1st International conference on Mediterranean Geosciences Union, MedGU 2021 ; Conference date: 25-11-2021 Through 28-11-2021",

year = "2024",

doi = "10.1007/978-3-031-43218-7_45",

language = "English",

isbn = "9783031432170",

series = "Advances in Science, Technology and Innovation",

publisher = "Springer Nature",

pages = "193--196",

editor = "Attila {\c C}iner and Erg{\"u}ler, {Zeynal Abiddin} and Mourad Bezzeghoud and Mustafa Ustuner and Mehdi Eshagh and Hesham El-Askary and Arkoprovo Biswas and Luca Gasperini and Klaus-G{\"u}nter Hinzen and Murat Karakus and Cesare Comina and Ali Karrech and Alina Polonia and Chamin{\'e}, {Helder I.}",

booktitle = "Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3",

address = "Switzerland",

}

Hesham, A & Seker, DZ 2024, Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices. in A Çiner, ZA Ergüler, M Bezzeghoud, M Ustuner, M Eshagh, H El-Askary, A Biswas, L Gasperini, K-G Hinzen, M Karakus, C Comina, A Karrech, A Polonia & HI Chaminé (eds), Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3. Advances in Science, Technology and Innovation, Springer Nature, pp. 193-196, 1st International conference on Mediterranean Geosciences Union, MedGU 2021, Istanbul, Turkey, 25/11/21. https://doi.org/10.1007/978-3-031-43218-7_45

Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices. / Hesham, Anas; Seker, Dursun Zafer.
Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3. ed. / Attila Çiner; Zeynal Abiddin Ergüler; Mourad Bezzeghoud; Mustafa Ustuner; Mehdi Eshagh; Hesham El-Askary; Arkoprovo Biswas; Luca Gasperini; Klaus-Günter Hinzen; Murat Karakus; Cesare Comina; Ali Karrech; Alina Polonia; Helder I. Chaminé. Springer Nature, 2024. p. 193-196 (Advances in Science, Technology and Innovation).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices

AU - Hesham, Anas

AU - Seker, Dursun Zafer

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

PY - 2024

Y1 - 2024

N2 - Water bodies like rivers and lakes are essential human, animal, and vegetation resources. Acquiring and analyzing these data are essential for better water resources and environmental management. In recent decades, many methods have been used to extract water bodies from remote sensing images. A critical way was using water indices such as normalized difference water index (NDWI) and modified normalized difference water index (MNDWI). Furthermore, extracting small water bodies (such as creeks) from satellite images is challenging and needs high spatial resolution images. However, there are several difficulties in using water indices in high spatial resolution images because of the limited spectral resolution. Most high spatial resolution satellites lack SWIR, which is necessary to calculate MNDWI. But, WorldView-3 overcomes this problem by having NIR and SWIR bands with high spatial resolution. This study used a Kaggle dataset called DSTL, which includes WorldView-3 images for the same region (but the dataset's provider obscures the location of these images) with 0.31 m, 1.24 m, and 7.5 m spatial resolution at nadir for panchromatic, multi-spectral, and SWIR, respectively. A convolutional neural network (CNN) was proposed to improve water body extraction from satellite images. In this network, three indices (normalized difference vegetation index (NDVI), NDWI, and MNDWI) were input layers to U-Net architecture. To evaluate this approach, the results were compared with the same U-Net architecture using RGB and all WorldView-3 bands as input layers, SegNet, and FCN-8. The Jaccard index was calculated as 90.7%, 90%, 89.3%, 85%, and 80% for U-Net with three indices, RGB, all WorldView-3 bands, SegNet, and FCN-8, respectively. The results show that using indices as input layers to U-Net improves the water body extraction, which is necessary to be accurate in several hydrological and water resources studies.

AB - Water bodies like rivers and lakes are essential human, animal, and vegetation resources. Acquiring and analyzing these data are essential for better water resources and environmental management. In recent decades, many methods have been used to extract water bodies from remote sensing images. A critical way was using water indices such as normalized difference water index (NDWI) and modified normalized difference water index (MNDWI). Furthermore, extracting small water bodies (such as creeks) from satellite images is challenging and needs high spatial resolution images. However, there are several difficulties in using water indices in high spatial resolution images because of the limited spectral resolution. Most high spatial resolution satellites lack SWIR, which is necessary to calculate MNDWI. But, WorldView-3 overcomes this problem by having NIR and SWIR bands with high spatial resolution. This study used a Kaggle dataset called DSTL, which includes WorldView-3 images for the same region (but the dataset's provider obscures the location of these images) with 0.31 m, 1.24 m, and 7.5 m spatial resolution at nadir for panchromatic, multi-spectral, and SWIR, respectively. A convolutional neural network (CNN) was proposed to improve water body extraction from satellite images. In this network, three indices (normalized difference vegetation index (NDVI), NDWI, and MNDWI) were input layers to U-Net architecture. To evaluate this approach, the results were compared with the same U-Net architecture using RGB and all WorldView-3 bands as input layers, SegNet, and FCN-8. The Jaccard index was calculated as 90.7%, 90%, 89.3%, 85%, and 80% for U-Net with three indices, RGB, all WorldView-3 bands, SegNet, and FCN-8, respectively. The results show that using indices as input layers to U-Net improves the water body extraction, which is necessary to be accurate in several hydrological and water resources studies.

KW - Convolutional neural network

KW - Deep learning

KW - Remote sensing

KW - Water body extraction

KW - Water index

KW - WorldView-3

UR - http://www.scopus.com/inward/record.url?scp=85189312536&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-43218-7_45

DO - 10.1007/978-3-031-43218-7_45

M3 - Conference contribution

AN - SCOPUS:85189312536

SN - 9783031432170

T3 - Advances in Science, Technology and Innovation

SP - 193

EP - 196

BT - Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3

A2 - Çiner, Attila

A2 - Ergüler, Zeynal Abiddin

A2 - Bezzeghoud, Mourad

A2 - Ustuner, Mustafa

A2 - Eshagh, Mehdi

A2 - El-Askary, Hesham

A2 - Biswas, Arkoprovo

A2 - Gasperini, Luca

A2 - Hinzen, Klaus-Günter

A2 - Karakus, Murat

A2 - Comina, Cesare

A2 - Karrech, Ali

A2 - Polonia, Alina

A2 - Chaminé, Helder I.

PB - Springer Nature

T2 - 1st International conference on Mediterranean Geosciences Union, MedGU 2021

Y2 - 25 November 2021 through 28 November 2021

ER -

Hesham A, Seker DZ. Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices. In Çiner A, Ergüler ZA, Bezzeghoud M, Ustuner M, Eshagh M, El-Askary H, Biswas A, Gasperini L, Hinzen KG, Karakus M, Comina C, Karrech A, Polonia A, Chaminé HI, editors, Recent Research on Geotechnical Engineering, Remote Sensing, Geophysics and Earthquake Seismology - Proceedings of the 1st MedGU, Istanbul 2021 Volume 3. Springer Nature. 2024. p. 193-196. (Advances in Science, Technology and Innovation). doi: 10.1007/978-3-031-43218-7_45

Investigating Accurate Water Body Extraction from Satellite Imagery Using Convolutional Neural Network with Water Indices

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this