Publication Cover
Canadian Journal of Remote Sensing
Journal canadien de télédétection
Volume 48, 2022 - Issue 2
187
Views
0
CrossRef citations to date
0
Altmetric
Articles

The Integration of Multi-Source Remotely Sensed Data with Hierarchically Based Classification Approaches in Support of the Classification of Wetlands

L'intégration des données de télédétection multi-sources avec des approches de classification hiérarchique à l'appui de la classification des zones humides

&
Pages 158-181 | Received 05 Apr 2021, Accepted 05 Aug 2021, Published online: 13 Nov 2021
 

Abstract

Methodologies were developed to classify wetlands (Open Bog, Treed Bog, Open Fen, Treed Fen, and Swamps) from remotely sensed data using advanced classification algorithms through two hierarchical approaches. The data utilized included multispectral optical and thermal data (Landsat-5, and Landsat-8), radar imagery (Sentinel-1), and a digital elevation model. Goals were to determine the best way to combine imagery to classify wetlands through hierarchically based classification approaches to produce more accurate and efficient maps compared to standard classification. Algorithms used were Random Forest (RF), and Naïve Bayes. A hierarchically based RF classification methodology produced the most accurate classification result (91.94%). The hierarchically based approaches also improved classification accuracies for low-quality data, as defined through feature analysis, when compared to a nonhierarchical classifier. The hierarchical approaches also produced a significant increase in classification accuracy for the Naïve Bayes classifier versus the standard approach (∼12% increase) while not significantly increasing computation time – comparable in accuracy to the RF tests for around 20% the computational effort. Preselection of spectral bands, polarizations and other input parameters (Normalized Difference Vegetation Index, Normalized Difference Water Index, albedo, slope, etc.) using log-normal or RF variable importance analysis was very effective at identifying low-quality features and features which were of higher quality.

Résumé

Des méthodologies ont été développées pour classer les zones humides (tourbière ouverte, tourbière arborée, tourbière minérotrophe ouverte, tourbière minérotrophe et marécages) à partir de données de télédétection à l’aide d’algorithmes de classification avancés par le biais de deux approches hiérarchiques. Les données utilisées comprenaient des données optiques et thermiques multispectrales (Landsat-5 et Landsat-8), des images radar (Sentinel-1) et des modèles numériques d’élévation (MNT). Les objectifs étaient de déterminer la meilleure façon de combiner l’imagerie pour classer les zones humides, au moyen d’approches de classification hiérarchiques pour produire des cartes plus précises et plus efficaces par rapport à une classification standard. Les algorithmes utilisés étaient les Forêts aléatoires (RF) et la classification naïve bayésienne. La classification RF hiérarchique a produit le résultat de classification le plus précis (91.94 %). Les approches hiérarchiques ont amélioré la précision de la classification pour les données de faible qualité, telles que définies par l’analyze des paramètres d’entrée, par rapport à un algorithme non hiérarchique. Les approches hiérarchiques ont également produit une augmentation significative de la précision pour la classification naïve bayésienne par rapport à l’approche standard (∼ 12% d’augmentation) sans augmenter significativement le temps de calcul – comparable en précision aux tests RF pour environ 20% de l’effort de calcul. La présélection des bandes spectrales, polarisations et autres paramètres d’entrée (NDVI, NDWI, albédo, pente…) à l’aide d’une analyse de l’importance des variables Log-Normal ou RF a été très efficace pour identifier les paramètres de faible qualité et ceux qui étaient de meilleure qualité.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Author contributions

Conceptualization of the study was developed by Aaron Judah and Baoxin Hu. Methodology was developed by Aaron Judah and validated by Baoxin Hu. All software was written by Aaron Judah. Formal analysis and investigation was carried out by Aaron Judah and validated by Baoxin Hu. Collection of imagery and other necessary resources was carried out by Aaron Judah. Data curation was done by Aaron Judah and validated by Baoxin Hu. Writing – original draft preparation was by Aaron Judah. Writing – review and editing was done by Baoxin Hu. Visualization work was done by Aaron Judah. Baoxin Hu was the project supervisor, project administrator and acquired the necessary funding.

Data availability statement

The data that support the findings of this study are available from the corresponding author, upon reasonable request.

Additional information

Funding

Funding was provided from NSERC under grant # 2021-03624. The European Space Agency Sentinel-1 imagery, and the use of the PolSARPro software. Natural Resources Canada and The Government of Canada provided the Canadian Digital Elevation Model.

Log in via your institution

Log in to Taylor & Francis Online

PDF download + Online access
  • 48 hours access to article PDF & online version
  • Article PDF can be downloaded
  • Article PDF can be printed
USD 61.00 Add to cart
* Local tax will be added as applicable

Related Research

People also read lists articles that other readers of this article have read.

Recommended articles lists articles that we recommend and is powered by our AI driven recommendation engine.

Cited by lists all citing articles based on Crossref citations.
Articles with the Crossref icon will open in a new tab.