CaltechAUTHORS
  A Caltech Library Service

Deep Learning-Based Damage Mapping With InSAR Coherence Time Series

Stephenson, Oliver L. and Köhne, Tobias and Zhan, Eric and Cahill, Brent E. and Yun, Sang-Ho and Ross, Zachary E. and Simons, Mark (2021) Deep Learning-Based Damage Mapping With InSAR Coherence Time Series. IEEE Transactions on Geoscience and Remote Sensing . ISSN 0196-2892. doi:10.1109/tgrs.2021.3084209. (In Press) https://resolver.caltech.edu/CaltechAUTHORS:20210728-221752758

[img] PDF - Accepted Version
See Usage Policy.

10MB
[img] PDF - Submitted Version
See Usage Policy.

22MB

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20210728-221752758

Abstract

Satellite remote sensing is playing an increasing role in the rapid mapping of damage after natural disasters. In particular, synthetic aperture radar (SAR) can image the Earth's surface and map damage in all weather conditions, day and night. However, current SAR damage mapping methods struggle to separate damage from other changes in the Earth's surface. In this study, we propose a novel approach to damage mapping, combining deep learning with the full time history of SAR observations of an impacted region in order to detect anomalous variations in the Earth's surface properties due to a natural disaster. We quantify Earth surface change using time series of interferometric SAR coherence, then use a recurrent neural network (RNN) as a probabilistic anomaly detector on these coherence time series. The RNN is first trained on pre-event coherence time series, and then forecasts a probability distribution of the coherence between pre- and post-event SAR images. The difference between the forecast and observed co-event coherence provides a measure of confidence in the identification of damage. The method allows the user to choose a damage detection threshold that is customized for each location, based on the local behavior of coherence through time before the event. We apply this method to calculate estimates of damage for three earthquakes using multiyear time series of Sentinel-1 SAR acquisitions. Our approach shows good agreement with observed damage and quantitative improvement compared to using pre- to co-event coherence loss as a damage proxy.


Item Type:Article
Related URLs:
URLURL TypeDescription
https://doi.org/10.1109/tgrs.2021.3084209DOIArticle
https://arxiv.org/abs/2105.11544arXivDiscussion Paper
ORCID:
AuthorORCID
Stephenson, Oliver L.0000-0002-5509-090X
Köhne, Tobias0000-0002-8400-7255
Cahill, Brent E.0000-0002-7723-3369
Yun, Sang-Ho0000-0001-6952-6156
Ross, Zachary E.0000-0002-6343-8400
Simons, Mark0000-0003-1412-6395
Additional Information:© 2021 IEEE. Manuscript received October 22, 2020; revised February 18, 2021 and April 27, 2021; accepted May 10, 2021. This work was supported by the National Aeronautics and Space Administration (NASA) Applied Sciences Disasters Program and performed at the Jet Propulsion Laboratory, California Institute of Technology. SLC processing was performed using the InSAR Scientific Computing Environment (ISCE) software from JPL/Caltech [60]. This work contains modified Copernicus data from the Sentinel-1A and 1B satellites processed by the European Space Agency and downloaded from the Alaska Satellite Facility Distributed Active Archive Center. Plots were produced using Matplotlib [61], Cartopy [62], and generic mapping tool (GMT) [63]. The RNN was built and trained using PyTorch [64]. Shuttle radar topography mission (SRTM) version 3 was used for the Ridgecrest digital elevation model. The authors would like to thank Heresh Fattahi and Piyush Agram for producing the coregistered SLC stack for the Sarpol-e-Zahab region, and Eric Fielding for producing the Ridgecrest stack. The authors would also like to thank Prof. Richard Murray, Stuart Feldman, and Eric Schmidt for assisting with the early development of the work. The authors are also grateful to Prof. Mahdi Motagh and two anonymous reviewers for helpful comments on the manuscript.
Group:Seismological Laboratory
Funders:
Funding AgencyGrant Number
NASA/JPL/CaltechUNSPECIFIED
Subject Keywords:Damage mapping, interferometric synthetic aperture radar (InSAR), machine learning, natural hazards
DOI:10.1109/tgrs.2021.3084209
Record Number:CaltechAUTHORS:20210728-221752758
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20210728-221752758
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:110057
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:02 Aug 2021 17:46
Last Modified:02 Aug 2021 17:46

Repository Staff Only: item control page