A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data

The number of severe storm events has increased in recent decades due to climate change. These storms are one of the main causes for timber loss in European forests and damaged areas are prone to further degradation by, for example, bark beetle infestations. Usually, manual mapping of damaged areas...

Full description

Bibliographic Details
Main Authors:	Daniel Scharvogel, Melanie Brandmeier, Manuel Weis
Format:	Article
Language:	English
Published:	MDPI AG 2020-11-01
Series:	Forests
Subjects:	CNNs remote sensing windthrow forest Deep Learning GIS
Online Access:	https://www.mdpi.com/1999-4907/11/12/1239

id	doaj-91f4422371864b5a934b4765506f1d0c
record_format	Article
spelling	doaj-91f4422371864b5a934b4765506f1d0c2020-11-27T07:53:49ZengMDPI AGForests1999-49072020-11-01111239123910.3390/f11121239A Deep Learning Approach for Calamity Assessment Using Sentinel-2 DataDaniel Scharvogel0Melanie Brandmeier1Manuel Weis2Esri Deutschland, Department Science and Education, Ringstr. 7, 85402 Kranzberg, GermanyEsri Deutschland, Department Science and Education, Ringstr. 7, 85402 Kranzberg, GermanyHessenForst, Europastraße 10-12, 35394 Gießen, GermanyThe number of severe storm events has increased in recent decades due to climate change. These storms are one of the main causes for timber loss in European forests and damaged areas are prone to further degradation by, for example, bark beetle infestations. Usually, manual mapping of damaged areas based on aerial photographs is conducted by forest departments. This is very time-consuming and therefore automatic detection of windthrows based on active and passive remote sensing data is an ongoing research topic. In this study we evaluated state-of-the-art Convolutional Neural Networks (CNNs) in combination with Geographic Information Systems (GIS) for calamity assessment. The study area is in in the northern part of Hesse (Germany) and was covered by twelve Sentinel-2 scenes from 2018. Labels of damaged areas from the Friedericke storm (18 January 2018) were provided by HessenForst. We conducted several experiments based on a custom U-Net setup to derive the optimal architecture and input data as well as to assess the transferability of the model. Results highlight the possibility to detect damaged forest areas using Sentinel-2 data. Using a binary classification, accuracies of more than 92% were achieved with an Intersection over Union (IoU) score of 46.6%. The proposed workflow was integrated into ArcGIS and is suitable for fast detection of damaged areas directly after a storm and for disaster management but is limited by the deca-meter spatial resolution of the Sentinel-2 data.https://www.mdpi.com/1999-4907/11/12/1239CNNsremote sensingwindthrowforestDeep LearningGIS
collection	DOAJ
language	English
format	Article
sources	DOAJ
author	Daniel Scharvogel Melanie Brandmeier Manuel Weis
spellingShingle	Daniel Scharvogel Melanie Brandmeier Manuel Weis A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data Forests CNNs remote sensing windthrow forest Deep Learning GIS
author_facet	Daniel Scharvogel Melanie Brandmeier Manuel Weis
author_sort	Daniel Scharvogel
title	A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data
title_short	A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data
title_full	A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data
title_fullStr	A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data
title_full_unstemmed	A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data
title_sort	deep learning approach for calamity assessment using sentinel-2 data
publisher	MDPI AG
series	Forests
issn	1999-4907
publishDate	2020-11-01
description	The number of severe storm events has increased in recent decades due to climate change. These storms are one of the main causes for timber loss in European forests and damaged areas are prone to further degradation by, for example, bark beetle infestations. Usually, manual mapping of damaged areas based on aerial photographs is conducted by forest departments. This is very time-consuming and therefore automatic detection of windthrows based on active and passive remote sensing data is an ongoing research topic. In this study we evaluated state-of-the-art Convolutional Neural Networks (CNNs) in combination with Geographic Information Systems (GIS) for calamity assessment. The study area is in in the northern part of Hesse (Germany) and was covered by twelve Sentinel-2 scenes from 2018. Labels of damaged areas from the Friedericke storm (18 January 2018) were provided by HessenForst. We conducted several experiments based on a custom U-Net setup to derive the optimal architecture and input data as well as to assess the transferability of the model. Results highlight the possibility to detect damaged forest areas using Sentinel-2 data. Using a binary classification, accuracies of more than 92% were achieved with an Intersection over Union (IoU) score of 46.6%. The proposed workflow was integrated into ArcGIS and is suitable for fast detection of damaged areas directly after a storm and for disaster management but is limited by the deca-meter spatial resolution of the Sentinel-2 data.
topic	CNNs remote sensing windthrow forest Deep Learning GIS
url	https://www.mdpi.com/1999-4907/11/12/1239
work_keys_str_mv	AT danielscharvogel adeeplearningapproachforcalamityassessmentusingsentinel2data AT melaniebrandmeier adeeplearningapproachforcalamityassessmentusingsentinel2data AT manuelweis adeeplearningapproachforcalamityassessmentusingsentinel2data AT danielscharvogel deeplearningapproachforcalamityassessmentusingsentinel2data AT melaniebrandmeier deeplearningapproachforcalamityassessmentusingsentinel2data AT manuelweis deeplearningapproachforcalamityassessmentusingsentinel2data
_version_	1724414158161575936

A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data

Similar Items