This paper aims at exploiting the potential of Global Navigation Satellite System Reflectometry (GNSS-R) for the detection of forest disturbances due to fires. The study focuses on the forested part of Angola that was largely affected by fires during summer 2019. The data collected in the area by the NASA Cyclone GNSS (CyGNSS) constellation have been downloaded and processed. As reference data for developing and testing the retrieval algorithms, the ESA Climate Change Initiative (CCI) decadal burned areas maps have been considered. After evaluating the sensitivity of the GNSS-R observables, namely Signal to Noise Ratio and Equivalent Reflectivity, to the forest disturbances, some retrieval algorithms based on different machine learning techniques have been implemented, tested, and compared with a simple approach based on the temporal gradient of the GNSS-R observables. In details, classifiers based on Support Vector Machines (SVM) and Random Forests (RF) have been implemented for discriminating burned/not burned pixels of the images and regressors based on Artificial Neural Networks (ANN), RF and Support Vector Regressors (SVR) have been implemented for estimating the fraction of burned areas within a pixel. All these algorithms allowed a satisfactory identification of the burned areas with respect to the reference data, enabling the generation of maps every ten days. These results represent the first observations of forest disturbances due to fires using spaceborne GNSS-Reflectometry.

Detecting fire disturbances in forests by using GNSS reflectometry and machine learning: A case study in Angola

Santi E.
;
2022

Abstract

This paper aims at exploiting the potential of Global Navigation Satellite System Reflectometry (GNSS-R) for the detection of forest disturbances due to fires. The study focuses on the forested part of Angola that was largely affected by fires during summer 2019. The data collected in the area by the NASA Cyclone GNSS (CyGNSS) constellation have been downloaded and processed. As reference data for developing and testing the retrieval algorithms, the ESA Climate Change Initiative (CCI) decadal burned areas maps have been considered. After evaluating the sensitivity of the GNSS-R observables, namely Signal to Noise Ratio and Equivalent Reflectivity, to the forest disturbances, some retrieval algorithms based on different machine learning techniques have been implemented, tested, and compared with a simple approach based on the temporal gradient of the GNSS-R observables. In details, classifiers based on Support Vector Machines (SVM) and Random Forests (RF) have been implemented for discriminating burned/not burned pixels of the images and regressors based on Artificial Neural Networks (ANN), RF and Support Vector Regressors (SVR) have been implemented for estimating the fraction of burned areas within a pixel. All these algorithms allowed a satisfactory identification of the burned areas with respect to the reference data, enabling the generation of maps every ten days. These results represent the first observations of forest disturbances due to fires using spaceborne GNSS-Reflectometry.
2022
Istituto di Fisica Applicata - IFAC
GNSS Reflectometry
CyGNSS
fire disturbances
File in questo prodotto:
File Dimensione Formato  
Detecting fire disturbances in forests by using GNSS reflectometry RED.pdf

solo utenti autorizzati

Tipologia: Versione Editoriale (PDF)
Licenza: NON PUBBLICO - Accesso privato/ristretto
Dimensione 1.37 MB
Formato Adobe PDF
1.37 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
Detecting fire disturbances in forests by using GNSS reflectometry _Elsevier_OA_POSTPRINT.pdf

accesso aperto

Descrizione: This is the Author Accepted Manuscript (postprint) version of the following paper: Santi E., Clarizia M. P., Comite D., Dente L., Guerriero L., Pierdicca N., "Detecting fire disturbances in forests by using GNSS reflectometry and machine learning: A case study in Angola", in REMOTE SENSING OF ENVIRONMENT, vol. 270, 2022, https://dx.doi.org/10.1016/j.rse.2021.112878
Tipologia: Documento in Post-print
Licenza: Creative commons
Dimensione 8.54 MB
Formato Adobe PDF
8.54 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.14243/441904
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 13
  • ???jsp.display-item.citation.isi??? 10
social impact