Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data

Shaojia Ge; Hong Gu; Weimin Su; Jaan Praks; Anne Lonnqvist; Oleg Antropov

doi:10.1109/IGARSS46834.2022.9883740

Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data

Shaojia Ge, Hong Gu, Weimin Su, Jaan Praks, Anne Lonnqvist, Oleg Antropov

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

Abstract

In this study, we evaluate the potential of deep learning models in predicting forest tree height in boreal forest zone using ESA Sentinel-1 and Sentinel-2 images. The performance of studied deep learning models is compared to several popular conventional machine learning approaches. The study area is located near Hyytiala forestry station in Finland, and represents a conifer-dominated mixed boreal forestland. Improved predictions were obtained when using combined optical and SAR data for all studied models. Our results indicate that UNet based models can achieve better accuracy in predicting forest tree heights (RMSE of 1.90m, mathrm{R}{2} of 0.69), compared to traditional parametric and machine learning models with RMSE range of 2.27-2.41m and mathrm{R}{2} range of 0.50-0.56 when satellite optical and radar data are combined.

Original language	English
Title of host publication	IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium
Publisher	IEEE
Pages	5688-5691
Number of pages	4
ISBN (Electronic)	978-1-6654-2792-0
DOIs	https://doi.org/10.1109/IGARSS46834.2022.9883740
Publication status	Published - 2022
MoE publication type	A4 Conference publication
Event	IEEE International Geoscience and Remote Sensing Symposium - Kuala Lumpur, Malaysia Duration: 17 Jul 2022 → 22 Jul 2022

Publication series

Name	IEEE International Geoscience and Remote Sensing Symposium proceedings
ISSN (Print)	2153-6996
ISSN (Electronic)	2153-7003

Conference

Conference	IEEE International Geoscience and Remote Sensing Symposium
Abbreviated title	IGARSS
Country/Territory	Malaysia
City	Kuala Lumpur
Period	17/07/2022 → 22/07/2022

Keywords

boreal
deep learning
forest height
forest management
Sentinel-1
Sentinel-2
synthetic aperture radar

Access to Document

10.1109/IGARSS46834.2022.9883740

Cite this

Ge, S., Gu, H., Su, W., Praks, J., Lonnqvist, A., & Antropov, O. (2022). Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data. In IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium (pp. 5688-5691). ( IEEE International Geoscience and Remote Sensing Symposium proceedings). IEEE. https://doi.org/10.1109/IGARSS46834.2022.9883740

@inproceedings{c1c9556ce6c04e2ca74a0bf80147a8ea,

title = "Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data",

abstract = "In this study, we evaluate the potential of deep learning models in predicting forest tree height in boreal forest zone using ESA Sentinel-1 and Sentinel-2 images. The performance of studied deep learning models is compared to several popular conventional machine learning approaches. The study area is located near Hyytiala forestry station in Finland, and represents a conifer-dominated mixed boreal forestland. Improved predictions were obtained when using combined optical and SAR data for all studied models. Our results indicate that UNet based models can achieve better accuracy in predicting forest tree heights (RMSE of 1.90m, mathrm{R}{2} of 0.69), compared to traditional parametric and machine learning models with RMSE range of 2.27-2.41m and mathrm{R}{2} range of 0.50-0.56 when satellite optical and radar data are combined.",

keywords = "boreal, deep learning, forest height, forest management, Sentinel-1, Sentinel-2, synthetic aperture radar",

author = "Shaojia Ge and Hong Gu and Weimin Su and Jaan Praks and Anne Lonnqvist and Oleg Antropov",

note = "Funding Information: The research was funded by Business Finland project MUL-TICO (Autonomous Sensing using Satellites, Multicopters, Sensors and Actuators) under grant agreement 6501/31/2019. And it was also supported by National Natural Science Foundation of China (Grant No. 61801221, 62001229), and by China Postdoctoral Science Foundation (Grant No. 2020M681604). Publisher Copyright: {\textcopyright} 2022 IEEE.; IEEE International Geoscience and Remote Sensing Symposium , IGARSS ; Conference date: 17-07-2022 Through 22-07-2022",

year = "2022",

doi = "10.1109/IGARSS46834.2022.9883740",

language = "English",

series = " IEEE International Geoscience and Remote Sensing Symposium proceedings",

publisher = "IEEE",

pages = "5688--5691",

booktitle = "IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium",

address = "United States",

}

Ge, S, Gu, H, Su, W, Praks, J, Lonnqvist, A & Antropov, O 2022, Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data. in IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium. IEEE International Geoscience and Remote Sensing Symposium proceedings, IEEE, pp. 5688-5691, IEEE International Geoscience and Remote Sensing Symposium , Kuala Lumpur, Malaysia, 17/07/2022. https://doi.org/10.1109/IGARSS46834.2022.9883740

Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data. / Ge, Shaojia; Gu, Hong; Su, Weimin et al.
IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2022. p. 5688-5691 ( IEEE International Geoscience and Remote Sensing Symposium proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data

AU - Ge, Shaojia

AU - Gu, Hong

AU - Su, Weimin

AU - Praks, Jaan

AU - Lonnqvist, Anne

AU - Antropov, Oleg

N1 - Funding Information: The research was funded by Business Finland project MUL-TICO (Autonomous Sensing using Satellites, Multicopters, Sensors and Actuators) under grant agreement 6501/31/2019. And it was also supported by National Natural Science Foundation of China (Grant No. 61801221, 62001229), and by China Postdoctoral Science Foundation (Grant No. 2020M681604). Publisher Copyright: © 2022 IEEE.

PY - 2022

Y1 - 2022

N2 - In this study, we evaluate the potential of deep learning models in predicting forest tree height in boreal forest zone using ESA Sentinel-1 and Sentinel-2 images. The performance of studied deep learning models is compared to several popular conventional machine learning approaches. The study area is located near Hyytiala forestry station in Finland, and represents a conifer-dominated mixed boreal forestland. Improved predictions were obtained when using combined optical and SAR data for all studied models. Our results indicate that UNet based models can achieve better accuracy in predicting forest tree heights (RMSE of 1.90m, mathrm{R}{2} of 0.69), compared to traditional parametric and machine learning models with RMSE range of 2.27-2.41m and mathrm{R}{2} range of 0.50-0.56 when satellite optical and radar data are combined.

AB - In this study, we evaluate the potential of deep learning models in predicting forest tree height in boreal forest zone using ESA Sentinel-1 and Sentinel-2 images. The performance of studied deep learning models is compared to several popular conventional machine learning approaches. The study area is located near Hyytiala forestry station in Finland, and represents a conifer-dominated mixed boreal forestland. Improved predictions were obtained when using combined optical and SAR data for all studied models. Our results indicate that UNet based models can achieve better accuracy in predicting forest tree heights (RMSE of 1.90m, mathrm{R}{2} of 0.69), compared to traditional parametric and machine learning models with RMSE range of 2.27-2.41m and mathrm{R}{2} range of 0.50-0.56 when satellite optical and radar data are combined.

KW - boreal

KW - deep learning

KW - forest height

KW - forest management

KW - Sentinel-1

KW - Sentinel-2

KW - synthetic aperture radar

UR - http://www.scopus.com/inward/record.url?scp=85140365372&partnerID=8YFLogxK

U2 - 10.1109/IGARSS46834.2022.9883740

DO - 10.1109/IGARSS46834.2022.9883740

M3 - Conference article in proceedings

AN - SCOPUS:85140365372

T3 - IEEE International Geoscience and Remote Sensing Symposium proceedings

SP - 5688

EP - 5691

BT - IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium

PB - IEEE

T2 - IEEE International Geoscience and Remote Sensing Symposium

Y2 - 17 July 2022 through 22 July 2022

ER -

Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

MULTICO: Autonomous Sensing using Satellites, Multicopters, Sensors and Actuators

Cite this

Deep Learning Models in Forest Mapping Using Multitemporal SAR and Optical Satellite Data

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Projects

MULTICO: Autonomous Sensing using Satellites, Multicopters, Sensors and Actuators

Cite this