Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge

Oluwanisola Ibikunle; Hara Madhav Talasila; Debvrat Varshney; John D. Paden; Jilu Li; Maryam Rahnemoonfar

doi:10.1109/RadarConf2351548.2023.10149734

Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge

Oluwanisola Ibikunle, Hara Madhav Talasila, Debvrat Varshney, John D. Paden, Jilu Li, Maryam Rahnemoonfar

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

Abstract

This paper documents the performance of two deep learning models developed to automatically track internal layers in Snow Radar echograms. A novel iterative RowBlock approach is developed to circumvent the small training-data problem peculiar to radar data by recasting pixel-wise dense prediction problem as a multi-class classification task with millions of training data. The proposed models, Skip_MLP and LSTM_PE, achieved tracking accuracies of 81.2 % and 87.9%, respectively, on echograms from the dry snow zone in Greenland. Moreover, 96.7% and 97.3% of the errors are less than or equal to two pixels for both models respectively. The tracked layers were used to estimate annual accumulation over two decades and compared with Regional Atmosphere Model (MAR) estimates to yield a coefficient of determination of 0.943, thus validating this approach.

Original language	English
Title of host publication	RadarConf23 - 2023 IEEE Radar Conference, Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781665436694
DOIs	https://doi.org/10.1109/RadarConf2351548.2023.10149734
State	Published - 2023
Externally published	Yes
Event	2023 IEEE Radar Conference, RadarConf23 - San Antonia, United States Duration: May 1 2023 → May 5 2023

Publication series

Name	Proceedings of the IEEE Radar Conference
Volume	2023-May
ISSN (Print)	1097-5764
ISSN (Electronic)	2375-5318

Conference

Conference	2023 IEEE Radar Conference, RadarConf23
Country/Territory	United States
City	San Antonia
Period	05/1/23 → 05/5/23

Funding

V. ACKNOWLEDGMENT We acknowledge the use of the CReSIS toolbox, data and/or data products from CReSIS generated with support from the University of Kansas, NASA Operation IceBridge grant NNX16AH54G, and NSF grants ACI-1443054, OPP-1739003, and IIS-1838230. Center for Remote Sensing and Integrated Systems (CReSIS) is the new name of the Center for Remote Sensing and Ice Sheets (CReSIS). This work is supported by NSF IIS-1838236 and ACI-1443054. Center for Remote Sensing and Integrated Systems (CReSIS) is the new name of the Center for Remote Sensing and Ice Sheets (CReSIS). This work is supported by NSF IIS-1838236 and ACI-1443054. We acknowledge the use of the CReSIS toolbox, data and/or data products from CReSIS generated with support from the University of Kansas, NASA Operation IceBridge grant NNX16AH54G, and NSF grants ACI-1443054, OPP- 1739003, and IIS-1838230.

Keywords

echogram
layer tracking
LSTM_PE
machine learning
multi-class classification
neural network
ResNet
Skip_MLP
snow radar

Access to Document

10.1109/RadarConf2351548.2023.10149734

Cite this

Ibikunle, O., Talasila, H. M., Varshney, D., Paden, J. D., Li, J., & Rahnemoonfar, M. (2023). Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge. In RadarConf23 - 2023 IEEE Radar Conference, Proceedings (Proceedings of the IEEE Radar Conference; Vol. 2023-May). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/RadarConf2351548.2023.10149734

@inproceedings{8dce75838c5a4000b690a3d78a6edefb,

title = "Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge",

abstract = "This paper documents the performance of two deep learning models developed to automatically track internal layers in Snow Radar echograms. A novel iterative RowBlock approach is developed to circumvent the small training-data problem peculiar to radar data by recasting pixel-wise dense prediction problem as a multi-class classification task with millions of training data. The proposed models, Skip_MLP and LSTM_PE, achieved tracking accuracies of 81.2 % and 87.9%, respectively, on echograms from the dry snow zone in Greenland. Moreover, 96.7% and 97.3% of the errors are less than or equal to two pixels for both models respectively. The tracked layers were used to estimate annual accumulation over two decades and compared with Regional Atmosphere Model (MAR) estimates to yield a coefficient of determination of 0.943, thus validating this approach.",

keywords = "echogram, layer tracking, LSTM_PE, machine learning, multi-class classification, neural network, ResNet, Skip_MLP, snow radar",

author = "Oluwanisola Ibikunle and Talasila, {Hara Madhav} and Debvrat Varshney and Paden, {John D.} and Jilu Li and Maryam Rahnemoonfar",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 IEEE Radar Conference, RadarConf23 ; Conference date: 01-05-2023 Through 05-05-2023",

year = "2023",

doi = "10.1109/RadarConf2351548.2023.10149734",

language = "English",

series = "Proceedings of the IEEE Radar Conference",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "RadarConf23 - 2023 IEEE Radar Conference, Proceedings",

}

Ibikunle, O, Talasila, HM, Varshney, D, Paden, JD, Li, J & Rahnemoonfar, M 2023, Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge. in RadarConf23 - 2023 IEEE Radar Conference, Proceedings. Proceedings of the IEEE Radar Conference, vol. 2023-May, Institute of Electrical and Electronics Engineers Inc., 2023 IEEE Radar Conference, RadarConf23, San Antonia, United States, 05/1/23. https://doi.org/10.1109/RadarConf2351548.2023.10149734

Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge. / Ibikunle, Oluwanisola; Talasila, Hara Madhav; Varshney, Debvrat et al.
RadarConf23 - 2023 IEEE Radar Conference, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2023. (Proceedings of the IEEE Radar Conference; Vol. 2023-May).

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

TY - GEN

T1 - Snow Radar Echogram Layer Tracker

T2 - 2023 IEEE Radar Conference, RadarConf23

AU - Ibikunle, Oluwanisola

AU - Talasila, Hara Madhav

AU - Varshney, Debvrat

AU - Paden, John D.

AU - Li, Jilu

AU - Rahnemoonfar, Maryam

PY - 2023

Y1 - 2023

N2 - This paper documents the performance of two deep learning models developed to automatically track internal layers in Snow Radar echograms. A novel iterative RowBlock approach is developed to circumvent the small training-data problem peculiar to radar data by recasting pixel-wise dense prediction problem as a multi-class classification task with millions of training data. The proposed models, Skip_MLP and LSTM_PE, achieved tracking accuracies of 81.2 % and 87.9%, respectively, on echograms from the dry snow zone in Greenland. Moreover, 96.7% and 97.3% of the errors are less than or equal to two pixels for both models respectively. The tracked layers were used to estimate annual accumulation over two decades and compared with Regional Atmosphere Model (MAR) estimates to yield a coefficient of determination of 0.943, thus validating this approach.

AB - This paper documents the performance of two deep learning models developed to automatically track internal layers in Snow Radar echograms. A novel iterative RowBlock approach is developed to circumvent the small training-data problem peculiar to radar data by recasting pixel-wise dense prediction problem as a multi-class classification task with millions of training data. The proposed models, Skip_MLP and LSTM_PE, achieved tracking accuracies of 81.2 % and 87.9%, respectively, on echograms from the dry snow zone in Greenland. Moreover, 96.7% and 97.3% of the errors are less than or equal to two pixels for both models respectively. The tracked layers were used to estimate annual accumulation over two decades and compared with Regional Atmosphere Model (MAR) estimates to yield a coefficient of determination of 0.943, thus validating this approach.

KW - echogram

KW - layer tracking

KW - LSTM_PE

KW - machine learning

KW - multi-class classification

KW - neural network

KW - ResNet

KW - Skip_MLP

KW - snow radar

UR - http://www.scopus.com/inward/record.url?scp=85163756798&partnerID=8YFLogxK

U2 - 10.1109/RadarConf2351548.2023.10149734

DO - 10.1109/RadarConf2351548.2023.10149734

M3 - Conference contribution

AN - SCOPUS:85163756798

T3 - Proceedings of the IEEE Radar Conference

BT - RadarConf23 - 2023 IEEE Radar Conference, Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 1 May 2023 through 5 May 2023

ER -

Ibikunle O, Talasila HM, Varshney D, Paden JD, Li J, Rahnemoonfar M. Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge. In RadarConf23 - 2023 IEEE Radar Conference, Proceedings. Institute of Electrical and Electronics Engineers Inc. 2023. (Proceedings of the IEEE Radar Conference). doi: 10.1109/RadarConf2351548.2023.10149734

Snow Radar Echogram Layer Tracker: Deep Neural Networks for radar data from NASA Operation IceBridge

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this